First a summary of what pollsters expect.

Anyone who was traumatized by 2024 - don’t trust the polls. We will know the answer one way or another soon enough. If the polls are right, likely hear Dems got the House even well before midnight on the Tuesday or else by midday Wednesday.

The pollsters are ALL projecting a similar majority in the House for Dems to their majority in 2018. Then they declared very fast. In 2018 we learnt about the US House results

• first from Fox News at 9.33 pm EST (that was the time of day in New York and Washington DC (2.33 am UK time)

• NBC next at 10.22 pm (3.22 am UK time) z

[If you need a reminder of which way around it is, the sun sets in the UK while still high in the sky further to the West in the US so it’s UK is later than the US and it’s later by 5 hours]

QUOTE STARTS

At 9:33 p.m., anchor Bret Baier declared, “We are now ready to make one of the biggest calls of the night. The Fox News Decision Desk can now project the Democrats will take control of the House of Representatives for the first time in eight years.”

….

A few minutes after 10 p.m., with still no other networks making the call, Fox News politics editor Chris Stirewalt appeared on air to explain what happened.

“Why are we making that call?” Baier asked him.

“We are just that good,” Stirewalt replied, crediting the Fox News Voter Analysis system. He said, for example, that the close Kentucky sixth district race, ultimately won by the Republican in a squeaker, provided one crucial indicator. “That was a good one to watch. The very fact that that race was as close as it was told us and told me that the Democrats were getting the numbers they need. They have no business doing so well in a district like that.”

…

Finally, at 10:22 p.m., NBC News was next to call the House for the Democrats. The announcement immediately prompted chatter on the MSNBC set about the congressional oversight likely to face the Trump White House in 2019.

…

It was after 11 p.m. before CNN made its own official call, further validating Fox News’ conclusion. The latter network was also first to declare that Republicans would maintain control of the Senate, at 10:23 p.m., though that call was less controversial by that point in the night.

https://www.politico.com/story/2018/11/07/fox-news-house-democrats-2018-elections-midterms-967820

A few early results with Dems winning in places they could only win with a big swing in different parts of the US and they were sure they had a majority in the House.

This is long before we know who won which seat in most of the states. It’s based on a few key races early on in widely different parts of the country. Based on that they already have such a good idea for the midterms that they can project confidently for the entire country what the result will be even though they can’t predict any individual close race.

We could know the Senate quite soon too if the polls are correct.

On the other hand if the polls are wrong, then the Senate could easily not be known for weeks in worst case if it depends on Alaska or Maine. The House could be delayed for a week or more if it is a very close election.

There is a lot of activity on election night and many states declare their results pretty fast on the East coast.

If you want to avoid the excitement, you could just check last thing at night before you go to bed in case it is already declared - and then again the next morning.

Or else just wait until the next day or when everyone else tells you.

If you are in the UK unless you stay up late on election night you are not likely to know the results until the next day. The first Florida results come in 7.30 pm ET which is 12.30 am UK time.

By avoiding election night you skip all the speculation about close races and the results seeming to point one way and then the other.

This page lists when all the states close polls.

https://www.thegreenpapers.com/G26/closing.phtml?format=gc

This page shows which states have declared results every half hour during election night based on past examples

Voting has just started. But possibly about half the votes are already cast. 57 million votes according to NBC who have been tracking them.

https://www.nbcnews.com/politics/2026-elections/early-vote

We will NOT get all the Trump cases in the courts because he has no standing to sue. He is not on the ballot.

So - if you find election night stressful - the simplest solution is simply to do something else and tune out of social media and the news if you can for a single night.

Check the news just before you go to bed on election night and if they already know the results well and good. If not then off to bed.

By midday the next day you find out the result along with most other Americans as most don’t bother to stay up to the small hours to follow the results.

And then either you hear that the Dems got the House or Senate or both, or we are in for a long wait for the House or Senate.

Either way things are not that likely to move much more on Wednesday.

So now to a bit more detail how the results come in. All the times here are in East Coast Time following Votebeat.

Midnight ET is 5 am UK time. The polls close in Florida at 7 pm ET time. That is midnight UK time.

Florida is the first to declare. By 10 pm on election day we likely have a very good idea about the situation in Florida for almost all except the very closest races.

That’s 3 am UK time.

The big question in Florida is how well did the gerrymander work for the Republicans. They hope to gain 4 Republican seats but that assumes that Hispanics vote for Republicans similarly to their 2024 Trump vote and that may not be a correct assumption.

https://en.wikipedia.org/wiki/2026_Florida_redistricting#Partisan_breakdown_of_the_new_congressional_map

We also get to see if the Democrat Emily Gregory who won a special election in Trump’s home district in a surprise in March gets to keep her seat in the Florida House of Representatives.

https://www.bbc.co.uk/news/articles/c9qd7wyln2jo

It makes no difference to the House and of no great significance for the Florida house of representatives which is more than ⅔ R.

https://en.wikipedia.org/wiki/Florida_House_of_Representatives

But it will hit the news if she keeps it surely just because it is where Trump lives.

Then later in the night Texas also counts most of its vote. That is early morning around 4 am East coast. But unless it is close we may get an idea of whether Talarico got into the Senate which would be a big story if he does and he is favoured to get the seat at present. Especially big news because Trump got Paxton nominated with a last minute tweet after the Republican party had spent $100 million to try to get John Cornyn nominated as the Republican candidate in the primaries.

But neither of these are guaranteed. If either of those things happen it’s a pretty good night for Dems.

We also get a first idea of whether the Texas gerrymander worked for Republicans.

So those are some of the things that we may hear already on election night.

We hear the East coast results first for the most part.

VoiteBeat show it like this:

First results for Florida. 8.30 pm Eastern Time.

This is the status by 11 pm Eastern Time when some Americans are already off to bed.

Maine is the main anomaly - one that’s on the East coast but is slow to count - and has ranked choice which could lead to very long delays.

QUOTE STARTS

Maine elections are run on the municipal, not county, level, and many Maine towns are quite small and rural. About 90 of the state’s 486 municipalities hand-count their ballots, which can be a slow process. Others don’t report results at all, other than by physically delivering ballots and memory sticks to the state capital of Augusta two days after the election, as required by law. As a result, in the past, more than 20% of Maine’s total vote has gone uncounted on election night.

State election officials will start tabulating all those results shortly after they’re delivered and determine if any races need to go to ranked choice tabulations. We don’t know the exact date that the ranked choice process would occur, but Secretary of State Shenna Bellows told the Maine Morning Star that it probably wouldn’t be until at least Friday, Nov. 13. That’s because election officials have to meticulously check and double-check that every single ballot is properly logged before running the calculations, which themselves take only a few minutes. As in the past, the secretary of state will live-stream the whole process.

https://projects.votebeat.org/2026/when-to-expect-midterm-election-results-2026/

11 pm Central time.

11 pm Mountain time.

11 pm Pacific time.

9 am next day Eastern time.

And midday Eastern time.

https://projects.votebeat.org/2026/when-to-expect-midterm-election-results-2026/

In past elections when the House was very close then California has been amongst the last. It’s likely to be amongst the last this time too because of its slow vote counting of postal ballots.

But it matters much less for this election because of the gerrymandering.

It is very blue. But there are some Republican counties in California. Nearly all are now Safe R or Safe D with far more safe D than safe R. The gerrymander got rid of most of the lean or likely R or D..

Zoom in:

Almost all are safe D or safe R - that is what gerrymandering is intended to do - carve the Republicans into districts almost all Republican so that the remaining districts are mainly safe Ds.

The only ones left not listed as safe D or Safe R in the RacetotheWH list are these three.

https://www.racetothewh.com/house

Unless the race is so close it depends on 3 seats to tell who has the House then California is not likely to delay declaring who won the House.

This is why it doesn’t matter that California won’t be declaring its results for a long time - is usually late. Because there are unlikely to be many surprises there.

For the likely results see:

About this substack, contact me, tips etc

But if you are scared by Trump please don’t count on this. Extraordinary things do happen sometimes. This is a “hailstones in summer in the UK type upset if the Republicans do keep the house and Senate.

What we can be sure of is that the Republicans will blame Trump for their losses whether they just scrape through or lose one or both chambers. And a 50 - 50 Senate with rebellious senators would mean a difficult last 2 years of his presidency.

That’s because of Trump’s bizarre style of campaigning. He’s made it all about him and for his style of campaign it was completely logical to pitch in front of the botched repair job on the Reflecting pool as the last image he wants his party to have of him before they vote.

It doesn’t work. The polls go more and more against him. But he seems set in his ways that this is how he will campaign and nobody can change his approach in his party.

According to RacetotheWH, it is now only a 3.6% chance of Republicans keeping the House.

RacetotheWH is the only one that shares all the data so you can dig in as much as you like and they have similar projections to everyone else. Here are its projections on November 2 made into a projection with dots for seats .

House Projections for Mid-Terms - (based on Race to the WH: Nov 2, 1:12 PM ET)

From the rest of the page:

And

https://www.racetothewh.com/house

If this is correct - with 224 seats through to lean D - we may get a pretty good idea and even a media projected win of the House in the early hours of Wednesday morning - for instance perhaps after the Texas results are almost all counted at around 4 am Eastern time.

We get the first indications from Florida - most of their votes are completed by midnight Eastern time. That combined with early results from elsewhere may give a first indication.

But it CAN flip back and forth during the night. If you are easily scared the best thing may be to forget about all this if you8 can until midday Wednesday. By then if it is as big a win as in 2018 or larger as it looks to be, then we’ll know the Dems got the House. If it’s a narrow win or the Republicans somehow manage to keep it then we won’t know for a few days to a week or more.

Then - the Dems are strongly favoured to win the Senate. It’s no longer seen as a tossup. But the Republicans could keep it. Surely with a reduced majority.

Senate Projections for Mid-Terms - (based on Race to the WH: Oct 26 2026)

https://www.racetothewh.com/senate/26

NY Times about how hard polling is for the mid terms. Basically they have to use fudge factors because the Republicans don’t respond to pollsters nearly as much as Dems. If they use the polling methods that worked for 2018 and 2022 and 2020 they will get a far better result for Dems.

But to be careful they use the results that are accurate for 2024 and 2016, That still gets the Dems winning the House.

But they caution that there is no reason to expect that 2024 and 2016 are the furthest the pollsters can be out in the direction of missing a Republican win.

Ends:

QUOTE STARTS

This time? Well, look for yourself. Here’s what our final polls this year would show if we had used the weighting from prior cycles:

• Michigan: D+16 (today’s actual published result is D+11)

• Alaska: D+6 (actual D+1)

• Iowa: D+4 (actual D+1)

• Kansas: D+10 (actual D+5)

• Maine: D+5 (actual D+6)

• Ohio: D+11 (actual D+6)

• Texas: D+9 (actual D+7)

On average, that’s a difference of more than three points.

What does all of this mean for 2026? Well, you can choose your own adventure. More than anything, it illustrates how hard it is to predict how the polls might err.

Maybe it implies that we’ve adopted weighting techniques that can handle the problems that caused the Times/Siena poll to overestimate Democratic strength in the past, especially in white working-class states. If so, maybe we’re dead-on.

Or perhaps one could argue that the Times/Siena polls will underestimate Democrats. Our polls were accurate in 2018 and 2022, after all, so maybe one should trust the old approach in a midterm election. If so, maybe the Democrats will do even better than our polls suggest.

Or maybe the extraordinarily blue results on our old methodology and the increased effect of weighting suggest that nonresponse bias is worse than ever. If so, polls could seriously underestimate Republicans once again. There’s no reason to assume that 2016 or 2020 is the worst case scenario.

We’ll find out soon enough.

https://www.nytimes.com/2026/11/01/upshot/poll-times-siena-senate-midterms.html

RacetotheWH has it as a win of Dems by 37 seats in the House. But they have a just over 1 in 30 chance of Republicans gaining the House.

Repeating their projection here:

House Projections for Mid-Terms - (based on Race to the WH: Nov 2, 1:12 PM ET)

You can’t directly conclude how far their 3.6% is away in the other direction for the best for Democrats. But if it was a symmetrical distribution it would be a majority of 74 seats.

That is a very wide spread even at 3.6%

That’s the best we can say.

But if Republicans do get the House and Senate one thing is pretty sure, it will be a narrower majority in the Senate. And with a more rebellious Republican party.

And the Senate is much more hard to predict.

That is because only 35 change this time and most are safe R or D. So it depends on a handful of individual races.

NY Times makes it 7 races and and the Dems have to get 4 of them.

• Michigan, Kansas, Texas, Maine, Ohio and, more narrowly, Alaska and Iowa.

We may not hear the final results for Alaska and Maine for a long time.

But if the projections are correct we may not need to know those results. It might be 51 + ?0-2 for Alaska and Maine.

QUOTE STARTS

The party is ahead in all seven states likeliest to decide control of the Senate: Michigan, Kansas, Texas, Maine, Ohio and, more narrowly, Alaska and Iowa.

…

Based on New York Times/Siena polls of 610 voters in Alaska, 705 voters in Iowa, 736 voters in Kansas, 714 voters in Maine, 1,011 voters in Michigan, 717 voters in Ohio and 1,215 voters in Texas.

A substantial Democratic lead in Texas? Kansas? Anything like these results would have been unfathomable just a few months ago.

In today’s polarized era, there’s no precedent for a party to fare this well in so many states that are generally safe wins for the other side. The only plausible comparisons are the Democratic victories in 2006 and 2008, but the recent polls outpace even those. The average of all polls shows Democrats leading in the same states as the Times/Siena surveys, giving the party the edge in more solidly Republican Senate seats than it managed to flip in 2006 and 2008 combined.

https://www.nytimes.com/2026/11/01/upshot/poll-times-siena-senate-midterms.html

There RacetotheWH has Maine, Ohio, and Texas as lean D for the Senate. And Michigan as Likely D.

It has Alaska, and Kansas as tossup along with South Carolina.

If the Ds get all the RacetotheWH likely and Lean D in the Senate they get it with a majority of 2, one over what’s needed.

Leave out Maine and it is still 51 seats

These are the four main NY Times scenarios. None of them have the Republicans get the House. But with one of them we don’t hear for the House for up to a week. The Republicans can get the Senate if they are significantly underestimated.

Wave size: D+5 to D+7 popular vote (similar to 2010 or 2014).

Polling error: Democratic strength substantially overestimated (similar to 2016 or 2020).

Outcome: Republicans hold the Senate fairly easily. Democrats win only a narrow-to-modest House majority. That House victory may not be called for days or even a week, as Democrats wait for states like California to count their mail-in ballots.

…

Wave size: D+7 to D+9 (similar to 2006 or 2018).

Polling error: The polls are generally accurate, but Republicans are modestly underestimated — either nationally or in crucial red states. (Similar to 2018 or 2024.)

Outcome: A big blue wave that nonetheless leaves Republicans breathing a sigh of relief — and Democrats disappointed. The Senate only narrowly remains Republican, even though Democrats fare better in safely Republican states than either party has fared on opposing turf in decades. Democrats make significant gains in the House.

…

Wave size: D+8 to D+10 (similar to 2006 or 2018).

Polling error: Reasonably accurate, including in the key Senate battlegrounds. (Similar to 2006 or 2022.)

The outcome: Democrats win control of the Senate. They win an even larger House majority than they did in 2018.

…

Wave size: D+10 and higher (similar to 2008 or larger than any modern electoral victory).

Polling error: Democrats outperform the polls in the most closely watched states or nationwide. (Similar to 2012, 2022 or 2025.)

Outcome: A large Democratic majority in the House (maybe 250 seats), a comfortable one in the Senate (53 or more), and the beginning of a new chapter of American politics.

[That is a majority of 250 to 185 or 65 seats majority]

https://www.nytimes.com/2026/10/29/upshot/midterm-elections-senate-house-scenarios.html

If the Republicans do get the Senate and even the House it will be with a greatly reduced Senate majority, everyone agrees there, and with more rebelious Republicans.

And there is no possibility of Trump passing any major partisan legislation in the remaining 20 legislative days to Jan 03 because they have such a full timetable and very rebellious Republicans after the way Trump has been campaigning for the midterms. Irrespective of the results, ordinary Republican representatives will feel let down and any that lost their seats are sure to blame Trump for it.

So the one thing inevitable is a continuing weakening of Trump’s political power over Congress whatever happens. Due to him just not listening to the concerns of Republican legislators about the way he campaigned for the mid terms.

In my next post I’ll cover what to look out for on election night.

For easily scared or nervous people - then it may be best to skip election night.

Even if the pollsters are right - they will start to get clear ideas by midnight Eastern time ffbut just based on the results from Florida and a few early results from other states and that’s hard to extrapolate to the whole country.

Will get a much better idea with the results from Texas about 4 am. Eastern time.

But it can flip back and forth a lot during the night.

It will be much clearer by midday Wednesday. By then if the pollsters are right we already know that the Dems have got the House and may know for the Senate.

If they are wrong then we may be in for a very long wait of a week or more. Again nothing much will happen during that time.

We will NOT get all the Trump cases in the courts because he has no standing to sue. He is not on the ballot.

That is when we likely hear that he is off playing golf and not available for comment - not certain but it’s a common response for him.

I’ll also do a post about what to expect when on election night.

About this substack, contact me, tips etc

When they are panicking the most scared people I help need something they can respond to instantly. That is why I use these graphics. Less scared people don’t need them so much. It is just for humour and only for sharing with scared people in these blog posts.

I used to use copy / paste montages with paintshop pro but it takes far more time, and nowadays few people would find the results acceptable. I can’t afford a graphic artist. So it is AI or nothing. And they are not taking away jobs. They have numerous bizarre glitches no human could make.

As for electricity use it’s minimal compared to e.g. the electricity used by my laptop and monitor. And the many claims about AI data centers don’t survive fact check - please check my debunk about that. No - data centers will NOT turn dry areas into deserts - just a slight warming over the server itself like a smaller version of the city heat island effect - and other misconceptions about them

More details:

Why I sometimes use chatbots to make humorous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles

Here are the central points:

• The only reason I do it is as a way to get a simple message across to people who panic who relate to graphics far faster than text.

• I used to use Photoshop but AI generated work works better. I couldn’t afford to employ human artists to do it.

• It’s not because I think it is a work of art. It is full of flaws and doesn’t look like art. It’s just for fun.

• I use grok if I need to do images with Trump, Putin etc, because Perplexity AI and Chat GPT won’t do graphics with recognizable public figures.

• There is no way that this is taking away jobs as I would otherwise just need to do something else myself, can’t pay for someone else to do it. I am only doing it to share them with scared people to help them be less scared.

• There aren’t any copyright issues because they are just generic in appearance and don’t try to imitate anyone’s style. There are copyright issues with some ways of using AI to generate art but they don’t apply to things like this.

  • If you ask it to do a graphic in the style of a Roman statue or Van Gogh that’s obviously fine.

  • But if you ask it to do a graphic in the style of a modern artist without permission that is definitely wrong.

  • The generic type graphics they do by default don’t imitate any individual artist and those are the ones I use.

• What I did before was use Paintshop Pro (similar to using Photoshop) and I could do most of these graphics with Paintshop Pro but it would look clunky and out of date nowadays as not many people do that now, and not be as effective and take longer to do.

• The power requirements for the very occasional use of AI for images is a minute fraction of my power requirements for daily activities. On the net zero path all our electricity will come from renewables by 2050.

• Servers have to use water responsibly and in areas with shortages of water for cooling they have minimal or no water requirements

• The chatbot boom is not going to lead to a recession - its financed by big companies that can afford to spend billions of dollars a year with no returns indefinitely - and will likely soon break even, some of them already

On those last two points see

BLOG: How Chatbots are being improved to use energy more efficiently
— stories that claim they will use too much energy to be useful in the future miss out these details
You can read it here:
https://doomsdaydebunked.miraheze.org/wiki/How_Chatbots_are_being_improved_to_use_energy_more_efficiently_-_stories_that_claim_they_will_use_too_much_energy_to_be_useful_in_the_future_miss_out_these_details

No - data centers will NOT turn dry areas into deserts - just a slight warming over the server itself like a smaller version of the city heat island effect - and other misconceptions about them

On several other topics, see also my
https://ddebunked.org/d/999-ai-has-be-worrying-about-the-future/5?cleanskipx

They all expect the Republicans to lose the House. Then varied views about to what extent they have a chance to keep the Senate. They also hedge hugely because they haven't got a fudge free way to counteract the polling errors that happened in 2024 and the way they could be over compensating or undercompensating for the tendency for Republicans to not answer questions from pollsters.

And then the big unknown of enthusiasm and turnout.

Trump gave his final closing remarks before the mid term elections.

Prompt: Gently humorous editorial cartoon illustration, landscape format. Setting: the Lincoln Memorial Reflecting Pool in Washington DC at golden hour, the Washington Monument visible reflected in the long pool, Lincoln Memorial in the background. A cartoon caricature of Donald Trump (blond swept hair, long red tie, navy suit) stands at a podium with a “Midterms 2026” sign at the edge of the pool, gesturing passionately at the water. The pool shows comic signs of a botched renovation: a few patchy mismatched tiles, a small orange traffic cone floating, a lopsided “Under Repair” sign, a confused duck. Next to the podium is an easel with an architect’s sketch of an enormous triumphal arch towering over a tiny Arc de Triomphe for scale. A small crowd of bemused onlookers, one checking a wristwatch, one yawning. A big clock prop near the podium reads “Minute 23 of 60” with a tiny label “Midterms mentioned!”.

Bold title banner across the top in clear readable lettering: “Trump’s final pitch to voters before the mid terms”

Large speech bubble from Trump, readable text: “My friend did NOT botch the reflecting pool - it was vandals - and I’ll build a triumphal arch larger than the Arc de Triomphe”

Caption box at the bottom, readable text: “Mentioned mid terms briefly at 23 minutes into a one hour speech, then returned to the reflecting pool”

Style: warm, colourful, light satirical newspaper cartoon, affectionate rather than harsh.

https://www.perplexity.ai/search/61153123-d538-4825-b080-b11935abd485

Few would be surprised that Trump gave a long speech about the reflecting pool - it is a natural climax of his bizarre election campaign message too voters

Trump rambling for most of his speech about the reflecting pool in his campaign speeches is normal for him. It’s not special. :Choosing to give his final pitch in front of the reflecting pool was a natural choice for him given how many times he talks about it and the level of emphasis he has put on it throughout his mid terms campaign.

It makes no political sense. It is not something that energizes voters to vote for him. But he is not a natural campaigner and he just doesn’t understand politics.

He isn't mad. He just hasn't got a clue about campaigning. He's det in his ways a bit as some old people become and he thinks in terms of business deals and doesn't understand campaigning or appealing to voters.

The Republicans asked him not to talk about the Reflecting pool for the very reason that based on his campaign speeches so far they expected it to be a major part of his final pitch to voters.

Republicans asked Trump NOT to talk about his work on the reflecting pool and his planned bigger than the Arc de Triomphe Architectural project. But that was the main thing he talked about in a 1 hour speech.

And talking about how the damaged sheeting he claims is vandalism without any proof - everyone else says it was just a botched job.

And that was the main content of the speech. He covered the midterms in passing at 23 minutes in then returned to the reflecting pool. That was his choice of location for his final speech. He was right by the botched reflecting pool.

Not an aberration for Trump - this is what the whole campaign has been about - about him and why voters should vote for him because of his priorities

This is not an aberration the whole campaign has been pretty much the opposite of what ANY seasoned campaigner would do. He just DOES NOT HAVE A CLUE HOW TO CAMPAIGN.

He has told Republicans that this election is all about him . And so this is his closing speech about why they should vote for him to fix issues with the economy. Because the reflecting pool was the fault of vandals not Trump. So they can trust Trump to fix the economy just like they can trust him to fix the damage to the reflecting pool caused he claims by vandals.

Something like that may be how it played out in his mind.

He tells Americans to accept higher gasoline prices because the Iran war was needed.

Basically it’s all back to front to normal campaigning.

Instead of saying “We can fix the economy, this is how we’ll do it and why you can trust us and vote for us” it is “Trust me, I know what I am doing, it’s all about me and that is why you can be confident I’ll fix the economy”.

QUOTE STARTS

President Donald Trump began his final scheduled in-person speech before Election Day with lengthy remarks about the Lincoln Memorial Reflecting Pool.

Trump first mentioned the election more than 20 minutes into the speech, calling it “the most important midterm, I believe, in our lifetime.”

Later in the speech he continued discussing the pool, at which he spoke, as well as his push to erect a massive triumphal arch in Washington and his plans to build a promenade along the Potomac River.

...

The speech, which lasted nearly an hour, was expected to serve as Trump’s “closing argument” on the midterms before millions of voters head to the polls to determine which party will control the House and Senate in the next Congress.

But Trump kicked off his remarks with more than 20 minutes of commentary that focused largely on the Reflecting Pool, along with his administration’s work on refurbishing other Washington-area monuments and cracking down on crime in the nation’s capital.

...

Trump’s first explicit reference to the midterms came around the 23-minute mark.

https://www.cnbc.com/2026/11/02/trump-midterm-election-speech-reflecting-pool.html

I will go into the detailed projections in the next post though likely most of you know them already.

I’ll also do a post about what to expect when on election night.

About this substack, contact me, tips etc

The MOAB does produce a mushroom shaped cloud when it detonates but this is NOT a nuke.

Light-hearted humorous two-panel cartoon comic strip, landscape format, wide 16:9, bright cartoon style with exaggerated expressions.

Left panel: Interior view from the open rear cargo ramp of a large four-engine propeller MC-130 military transport aircraft flying at night over the desert mountains of Iran. Several cartoon soldiers in flight suits and helmets strain comically to push an enormous, oversized bomb (a MOAB, long cylindrical bomb with fins) strapped onto a rickety old wooden sled with curved runners out the back of the plane. Outside, surface-to-air missiles streak past the aircraft with smoke trails. One crew member shouts with a big speech bubble reading exactly: “Now let’s get home fast or we’ll be shot down!”

Right panel: A rugged rocky mountain (Pickaxe Mountain) under a dark sky, with a huge dramatic fiery orange mushroom-shaped explosion billowing over the top of the mountain. A cutaway cross-section below the mountain shows deep underground tunnels where small cartoon soldiers stand calmly, one holding a cup of tea, looking up puzzled as a little dust falls from the ceiling. Small “boom...” sound effect text faintly near the tunnel ceiling. Speech bubble from one tunnel soldier reading exactly: “Was that it?”

Clear readable comic lettering, panel border between the two panels.

https://www.perplexity.ai/search/6e526f9a-43ae-4141-bc3b-1ad1c2655ac5

Short note about my use of AI graphics

When they are panicking the most scared people I help need something they can respond to instantly. That is why I use these graphics. Less scared people don’t need them so much. It is just for humour and only for sharing with scared people in these blog posts.

I used to use copy / paste montages with paintshop pro but it takes far more time, and nowadays few people would find the results acceptable. I can’t afford a graphic artist. So it is AI or nothing. And they are not taking away jobs. They have numerous bizarre glitches no human could make.

As for electricity use it’s minimal compared to e.g. the electricity used by my laptop and monitor. And the many claims about AI data centers don’t survive fact check - please check my debunk about that. No - data centers will NOT turn dry areas into deserts - just a slight warming over the server itself like a smaller version of the city heat island effect - and other misconceptions about them

Robert's Short Doomsday Debunks

Why I sometimes use chatbots to make humerous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles…

Read more

7 months ago · 1 like · Robert Walker

The MOAB can’t be deployed from a B-52 stealth bomber it is also an extreme risk to the large, slow and obvious military transport plane that has to fly over Iran to drop it, very vulnerable to the Iranian air defences. And can only be dropped by pushing it out of the hold on a wooden sled with a drag parachute. And just a thin aluminium casing with zero penetration capability.

QUOTE STARTS

Unlike standard aerial bombs carried in the internal bomb bays of fighter jets or stealth bombers, the sheer bulk and length of the MOAB make it impossible to fit inside conventional combat aircraft.

Instead, it is loaded into the cargo hold of an MC-130 transport aircraft, pushed out on a wooden sled via a drag parachute, and guided toward its target using satellite positioning.
…
The MOAB is designed primarily as an air-burst weapon. Guided by satellite coordinates, the bomb does not impact the ground to create a deep crater. Instead, it uses a proximity fuze to detonate approximately 6 feet (1.8 metres) above the ground surface.
…

Despite being designed in 2002 and tested in 2003 at Eglin Air Force Base, Florida, the MOAB sat in the US arsenal unused for nearly 15 years.

The MOAB saw its combat debut on April 13, 2017, when a US Air Force MC-130 dropped a single GBU-43/B on an ISIS-K (Islamic State Khorasan Province) tunnel and cave complex in the Achin District of Nangarhar Province, Afghanistan.

The target was a dense complex of natural caves, improvised tunnels, and anti-personnel minefields used by ISIS fighters to evade air surveillance.

The strike collapsed cave entrances, detonated secondary munitions caches, cleared heavily mined terrain, and eliminated over 90 combatants without causing confirmed civilian casualties, according to official reports.

The 2017 attack demonstrated the exact role the MOAB was created for that is neutralising spread-out surface networks, shallow cave complexes, and clearing heavily defensive terrain where ground forces would otherwise face severe casualties.
…
The MOAB is almost entirely ineffective against deeply buried nuclear facilities such as Fordow or Natanz.

The zero penetration design of the MOAB is a thin aluminum casing built to maximise explosive blast energy in open air. If dropped directly onto solid rock or reinforced concrete, the bomb casing would break apart upon impact prior to optimal detonation. It cannot dig through rock.
…
One major concern is also the fact that the MOAB must be dropped from a slow-moving, non-stealthy cargo plane (MC-130).

Flying a cargo plane into heavily defended Iranian airspace — guarded by air defence systems like the S-300 and domestic Khordad-15 batteries — would expose the delivery crew to extreme risk unless total air supremacy is established first.

https://www.firstpost.com/explainers/what-is-america-mother-of-all-bombs-can-it-destroy-iran-buried-nuclear-sites-14053067.html

It’s of rather limited use which is why the US has used it only once. It is legal to use it in a military conflict so long as the use is proportional, not excessive risk to civilians etc, and it has not got the numerous problems of nukes, but it just isn’t of much military value in most situations.

It is not much use against Iran. It’s designed to cause a lot of destruction in a small area at ground level. It can also be used against bunkers but only if they are quite shallow as it is not a bunker buster. It was used against bunkers in Afghanistan.

Not likely to be any use against the very deeply buried most important bunkers in Iran. Also it can’t be deployed from the B-52 so it would need to be dropped from a non stealth bomber and that is a vulnerability.

If he used it, it would be a large spectacular explosion - especially if he used it above ground level - but it would be political theatre and not of much military value against Iran.

QUOTE STARTS

Brussels-based military expert Elijah Magnier told Al Jazeera that using the MOAB would not even guarantee the destruction of Iran’s deeply buried underground nuclear installations and that such a strike would likely amount to little more than “political theatre”.
…
Designed to explode a few feet above the ground, the MOAB sends out lethal waves of fire when it detonates, blasting away buildings, cave openings and other hard-surface structures.

Military expert Colonel Nidal Abu Zeid told Al Jazeera the bomb does not require impact in order to detonate and that its massive blast will destroy everything above ground within a 1km (0.6-mile) radius.
….

However, Magnier said the bomb’s radius and level of destruction would depend greatly on the terrain it is used within and how populated the area is. He added that it may not be particularly effective against targets buried deep underground.

Some of Iran’s nuclear sites which the US may wish to target, such as the Fordow complex and an under-construction site at the Pickaxe Mountain, are heavily fortified and deeply buried.

“Its large explosive charge can produce devastating blast effects against exposed personnel, equipment and vulnerable structures,” Magnier said. “The Afghanistan operation demonstrates its intended application against a cave-and-tunnel position; it does not demonstrate an ability to destroy every kind of underground installation.”

While its blast can reach underground, the MOAB is not especially designed to punch through deep concrete or mountain rock in the way that a “bunker buster” bomb is. The US military’s most powerful bunker buster is the GBU-57 Massive Ordnance Penetrator. Weighing about 30,000 pounds (13,600kg), including a 2,700kg (6,000 pound) warhead, this precision-guided bomb is made of high-strength steel and built to penetrate up to 200 feet (61 metres) underground before exploding.
…
However, besides not being guaranteed to hit underground targets effectively, Magnier told Al Jazeera it could be difficult to deploy the necessary aircraft in the regions of Iran where nuclear sites are located. Stealth bombers which are better suited to this terrain are not able to carry this size of bomb.

Using the bomb could also destroy infrastructure without inducing Iran to make any concessions, he added, noting that its deployment would risk turning “a spectacular event” into one that ultimately delivers few political results for Washington.

“The question I would put to any commander advocating its use is straightforward: What will this weapon accomplish that another available weapon cannot, and how will that result advance an achievable political settlement?,” Magnier said.

“If the answer centres on the spectacle of the explosion, the proposal is stronger as political theatre than as military strategy,” he added.

https://www.aljazeera.com/news/2026/10/15/what-is-the-mother-of-all-bombs-us-could-use-in-iran

See also our discussions on ddebunked.org debunking this:

[Answer: To use the MOAB would just be political theatre with little military value
— video contradicts itself and debunks itself in the first five minutes
— to post:] ◀ can he do this? (2026: November 1)
You can read it here:
https://ddebunked.org/d/6500-6500-answer-to-use-the-moab-would-just-be-political-theatre-with-little-military-value-video-contradicts-itself-and-debunks-itself-in-the-first-five-minutes-to-post-can-he-do-this

[Answer: The MOAB is CONVENTIONAL
— NOT A NUKE
— would be incredibly INEFFECTIVE against Iran's buried facilities
— only useful for surface explosions
— little more than political theatre if used
— to post:] ◀ They're gonna fucking use a nuke aren't they?.... (2026: October 16)
You can read it here:
https://ddebunked.org/d/6244-6244-answer-the-moab-is-conventional-not-a-nuke-would-be-incredibly-ineffective-against-irans-buried-facilities-only-useful-for-surface-explosions-little-more-than-political-theatre-if-used-to-post-theyre-gonna-fucking-use-a-nuke-arent-they

[Answer: MOAB would be incredibly ineffective against buried Iranian facilities
— achieves nothing except surface explosion as political theatre
— can only be pushed out of hold of a large slow obvious miltiary transport
— extreme risk to crew and aircraft over Iran
— with thin aluminium shell that can't penetrate rock
— to post:] ◀ MOAB being used and strikes before midterms? (2026: October 15)
You can read it here:
https://ddebunked.org/d/6218-6218-answer-moab-would-be-incredibly-ineffective-against-buried-iranian-facilities-achieves-nothing-except-surface-explosion-as-political-theatre-can-only-be-pushed-out-of-hold-of-a-large-slow-obvious-miltiary-transport-extreme-risk-to-crew-and-aircraft-over-i

Search here: Answered, Debunks and Positive News in ddebunked.org

See also:

About this substack, contact me, tips etc

I am not predicting here, I can’t see the future. But he has a strong pattern of responding to setbacks by playing golf and stopping press conferences. He’s done that several times since the start of his second term too.

Prompt:

Humorous editorial cartoon, split into two panels side by side. Top caption across the whole image in bold readable lettering: “IF TRUMP LOSES THE MIDTERMS...”

Left panel, smaller, labelled “WHAT YOU MIGHT EXPECT”: a cartoon giant fiery dragon with a recognisable Donald Trump-style blond swoop of hair, breathing flames over a Capitol building. A huge thick red X is drawn over the entire panel, crossing it out.

Right panel, larger, labelled “WHAT ACTUALLY HAPPENS”: a sunny, peaceful golf course. In the far distance, a tiny caricature of Donald Trump in a red cap and white polo shirt calmly swings a golf club. In the foreground, a crowd of press reporters and photographers with cameras, long lenses and notepads crouch behind a tall green hedge, peering through gaps and parting the leaves to get a glimpse. One reporter uses binoculars. Lighthearted, colourful, classic newspaper caricature style, gentle satire, clear composition.

https://www.perplexity.ai/search/ba290c56-2674-4d87-8972-30032332b70f

Short note about my use of AI graphics

The graphics are just for fun and I find these graphics are very useful in helping scared people, far faster in getting the message across than text.

They are only for sharing with scared readers of my post. They frequently have many obvious flaws for anyone with a keen eye. What I did before was use Paintshop Pro (similar to using Photoshop) but it would look clunky and out of date nowadays and take longer to do.

There aren’t any copyright issues with the way I use this, because they are just generic in appearance and don’t try to imitate anyone’s style. And I couldn’t afford to pay a professional illustrator. So they are not taking away jobs and as for electricity use it’s minimal compared to e.g. the electricity used by my laptop and monitor.

Robert's Short Doomsday Debunks

Why I sometimes use chatbots to make humerous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles…

Read more

7 months ago · 1 like · Robert Walker

And then longer term in 2020 he did of course try to stop certification of the results on Jan 06 but he FAILED.

But if you look at his situation today he has NOTHING he can do after the mid terms that could plausibly do anything.

And most people forget how in 2020 his first response was to retreat from public view and indeed all the way through to the end of his presidency he just stopped engaging nearly so much publicly after he lost to Biden and was spending lots of time secluded from the public.

We MAY be able to get a rough idea of how Trump will respond if he loses from what happened after he lost in 2020 to Biden. What he did is play lots of golf and not be seen for days.

https://www.bbc.co.uk/news/election-us-2020-54860213

As we all know he also launched lots of election challenges in 2020 and also tried to organize an overthrow of the electoral college ballot.

But he can’t do those things for the mid terms. There is no electoral college to overthrow.

The other big difference is that in 2024 Trump had standing as one of the candidates so he could take out legal cases against any of the counts in any of the states. In the midterms only the representatives voted for have standing. And Trump is not on the ballot.

So he can’t take out the same kinds of cases he took out in 2024. He can direct the DoJ to investigate some kinds of things but not challenge any of the election results himself. He can encourage Republican candidates to challenge them and provide funding for them to challenge them but that would be up to them not him and he is NOT popular with Republicans any more.

So we may see just the retreat to play golf and not much else.

There is nothing equivalent to Pence reading the ballot that he can overthrow, and no electoral college just the straight results with clear legal procedures.

The Federal government has limited ability to challenge the results.

And he’s already done all he can by way of executive orders.

He will have marginal control of Congress for 20 legislative days but with a very rebellious Republican party and a huge amount to get through in limited time - not much he can do there.

He has already purged everyone he can.

And he can purge cabinet members but the problem there is that he can only replace them by new cabinet members approved by the Senate and they are far less inclined to support unsuitable choices now.

So although there will likely be a lot of internal divisions in the cabinet and the executive and recriminations about why they lost - it’s not likely to lead to revolving door type firings because he can’t replace them with anyone better from his point of view.

Based on that I think the most likely outcome if he loses the House is he just retreats from public view and plays golf for at least a week or two., And prepares for likely impeachment proceedings in whatever way he can.

And - he may or may not try to wrap things up in the Iran war., He’ll be under lots of pressure to do so. And he may try to make progress with Ukraine and North Korea.

But generally he’ll be deflated and retire from public view.

And there is very little he can do in response in the case of the mid terms.

This is NOT prediction. He may not retreat to the golf courses.

He may stay in public view and bounce back in some stronger way than that.

But whatever it is - not likely to be the fire breathing dragon many of you worry about.

Anyway we’ll see what happens.

About this substack, contact me, tips etc

Our voluntary fact checkers on our new forum ddebunked.org or our Facebook Group "Doomsday Debunked" can help you if scared. Clickbait can impact mental health.

These posts are written by me, Robert, though they are the result of lots of collective searching for good material to help scared people on our forums.

I’m a mathematician and a music software developer, author of Bounce Metronome and Tune Smithy. I have a BSc in Maths (first class), and a MHum (a second undergraduate degree in philosophy done as a postgraduate course).

I studied for postgraduate at Oxford University but I didn’t complete my doctorate. my supervisor Robin Gandy advised me to write my work up as papers and get those published and apply for a DSc instead of a PhD but though I drafted out a paper, I have never tried to get it published.

However that gave me the background in checking sources and how to read academic papers which has been very useful for my science blogging and then for voluntary fact checking.

More about me here

https://robertinventor.online/booklets/about.htm

FOR MORE HELP

To find a debunk see: List of articles in my Debunking Doomsday blog to date See also my Short debunks

Scared and want a story debunked? Post to our forum which is set up for voluntary fact checking.

The forum itself is here:

ddebunked.org

Also do join our forum if you can help with fact checking or to help scared people who are panicking.

Alternatively you can post to our Facebook group. Please look over the group rules before posting or commenting as they help the group to run smoothly

• Facebook group Doomsday Debunked

However I am not able to comment or post there at present because

SEARCH LIST OF DEBUNKS

You can search by title and there’s also an option to search the content of the blog using a google search. Try different terms e.g. Russia, Putin etc as it only searches the title.

CLICK HERE TO SEARCH: List of articles in my Debunking Doomsday blog to date

You can also search our forum posts similarly

https://ddebunked.org/debunks.html

NEW SHORT DEBUNKS

I do many more fact checks and debunks on our Facebook group than I could ever write up as blog posts. They are shorter and less polished but there is a good chance you may find a short debunk for some recent concern.

I often write them up as “short debunks”

See Latest short debunks for new short debunks

I also tweet the debunks and short debunks to my Blue Sky page here:

. doomsdaydebunks.bsky.social

I do the short debunks more often but they are less polished - they are copies of my longer replies to scared people in the Facebook group.

I go through phases when I do lots of short debunks. Recently, I’ve taken to converting comments in the group into posts in the group that resemble short debunks and most of those haven’t yet been copied over to the wiki.

TIPS FOR DEALING WITH DOOMSDAY FEARS

If suicidal or helping someone suicidal see my:

BLOG: Supporting someone who is suicidal

If you have got scared by any of this, health professionals can help. Many of those affected do get help and find it makes a big difference.

They can’t do fact-checking, don’t expect that of them. But they can do a huge amount to help with the panic, anxiety, maladaptive responses to fear and so on.

Also do remember that therapy is not like physical medicine. The only way a therapist can diagnose or indeed treat you is by talking to you and listening to you. If this dialogue isn’t working for whatever reason, do remember you can always ask to change to another therapist and it doesn’t reflect badly on your current therapist to do this.

Also check out my Seven tips for dealing with doomsday fears based on things that help those scared, including a section about ways that health professionals can help you.

I know that sadly many of the people we help can’t access therapy for one reason or another - usually long waiting lists or the costs.

There is much you can do to help yourself. As well as those seven tips, see my:

BLOG: Doing what matters at times of stress
— new WHO booklet to help with mental health during COVID19 pandemic

BLOG: Breathe in and out slowly and deeply and other ways to calm a panic attack

BLOG: Tips from CBT
— might help some of you to deal with doomsday anxieties

PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

PLEASE DON’T COMMENT ON THIS POST WITH POTENTIALLY SCARY QUESTIONS ABOUT ANY OTHER TOPIC:

INSTEAD PLEASE COMMENT IN OUR NEW FORUM HERE:

ddebunked.org

It’s just a forum not social media. No age verification. Set up by myself with free open source software Flarum.

For details:

Doomsday Debunked - fact checking stories that scare you

Welcome to our own easy to use new discussion forum at ddebunked.org - voluntary fact checkers helping you if scared of many things - single click to join from Google, X or FB or join via email

We recently opened up our new dedicated Doomsday Debunked discussion forum just for voluntary fact checking for easily scared people. ddebunked.org You are all welcome to join us there…

Read more

a year ago · 3 likes · 1 comment · Robert Walker

If you have any issues joining it do let me know.

Why I ask you to post to our forum with anything scary off topic instead of here

The reason I ask you to post there instead of comment with your concerns about unrelated topics here is that I often can’t respond to comments here for some time. The unanswered comment can scare people who come to this post for help on something else

Also even an answered comment may scare them because they see the comment before my reply.

Do comment here with anything that is on topic for this post, E.g. if you spot any mistakes however small please let me know.

Also, though your first comment should be on topic, it is absolutely fine to digress and go off topic in conversations here in a natural way if that is how the conversation develops.

This is specifically about off topic comments here hat might scare people on a different topic.

PLEASE DON’T TELL A SCARED PERSON THAT THE THING THEY ARE SCARED OF IS TRUE WITHOUT A VERY RELIABLE SOURCE OR IF YOU ARE A VERY RELIABLE SOURCE YOURSELF - AND RESPOND WITH CARE

This is not like a typical post on substack. It is specifically to help people who are very scared with voluntary fact checking. Please no politically motivated exaggerations here. And please be careful, be aware of the context.

We have a rule in the Facebook group and it is the same here.

If you are scared and need help it is absolutely fine to comment about anything to do with the topic of the post that scares you.

But if you are not scared or don’t want help with my voluntary fact checking please don’t comment with any scary material.

If you respond to scared people here please be careful with your sources. Don’t tell them that something they are scared of is true without excellent reliable sources, or if you are a reliable source yourself.

It also matters a lot exactly HOW you respond. E.g. if someone is in an area with a potential for earthquakes there’s a big difference between a reply that talks about the largest earthquake that’s possible there even when based on reliable sources, and says nothing about how to protect themselves and the same reply with a summary and link to measures to take to protect yourself in an earthquake.

https://www.perplexity.ai/search/3758b300-f23a-4579-87b3-85eb0829b708

Robert's Short Doomsday Debunks

Why I sometimes use chatbots to make humerous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles…

Read more

7 months ago · 1 like · Robert Walker

Also

And then I have a suggestion for the future:

We get a surge of falsely scary gossip before every US elections - I think that the reason is likely that people have heightened anxiety and this reduces their discrimination. So they hit click and share with even less thought and discrimination than usual.

It is NOT that more things are happening.

For older people who remember the world before social media it might help to think of the social media feed as gossip. We are used to dismissing gossip.

Videos that look professional are often just flung together in seconds with modern tools

We are not yet used to dismissing videos with scary looking graphics and sometimes music and that may seem professionally made to less experienced viewers. Many of them can be built in a few minutes or seconds with simple apps and very little thought.

Things that only the big movie companies could do a couple of decades ago are now things almost anyone can do in an app of some sort. A simple example of that. Turning yourself visually into a cat for a video using a cat filter :). That was just about impossible 20 years ago say.

Never mind ACCIDENTALLY turning yourself into a cat in a serious video chat as a lawyer :).

That’s not very realistic by modern standards. As you likely know the more realistic cat filters show your eyes move and your facial expression change with the cat filter fitting exactly onto your face.

It’s very easy for people with the right apps to make all sorts of extraordinary reels that may seem believable especially to the more digitally naive viewers. And they may make them in seconds depending how they did it.

And many people can fabricate false stories just as fast as it takes to say them - with no extra thought whatsoever.

Social media is very new - and out of billions online - just takes one person to hit on the right combination of triggers for lots of people to hit and share - not because it is true - because it is FALSELY triggering

What’s happened is that social media is very new. And with billions of people online - then videos spread that were made by only one or two people out of a billion who happened to hit on just the right combination of ways to say things that made people click and share what they said.

These memes hijack the natural instincts and “gut feeling” of people that works very well in most situations. The emotions and feelings that lead you to act first and check later if someone shouts fire in a cinema - and the fire alarms go off - or if you see someone who is in serious trouble, a stranger that is at risk of dying. These emotions are very powerful and they override logic and calm reasoning any day.

So - they are good instincts you or your friends or colleagues or relatives or parents have that help protect you and them.

But they are co-opted by people who make these reels who use the exact same things that lead you to run out of a cinema if someone shouts fire, or to run to someone’s help if they are in serious trouble - to get you to do maladaptive things like prepping after watching a short reel by someone you know nothing about who is a complete stranger.

Only had falsely scary recommendations from strangers mixed into social media since 2020 - so it’s very new to us

And it’s because social media is so new. And mixing in stuff from random strangers in your news feed is even newer. We started to notice it in the group in a big way around 2020.

Humorous, light-hearted cartoon illustration in wide landscape format (16:9), split into two side-by-side panels comparing social media feeds. Clean, friendly flat cartoon style with soft colours.

Left panel, header text: “Before 2020: What you want to see”. Shows a smartphone-style social media feed with cosy posts from friends and family, each with a small friendly avatar: a birthday cake with candles captioned “Happy 40th, Sue!”, a child on a first day of school captioned “Look how tall Tom is now!”, friends laughing at a picnic captioned “Great day out with the gang”, a home-cooked lasagne captioned “Made this tonight!”. Warm, sunny, calm mood. A relaxed cartoon person scrolling and smiling.

Right panel, header text: “After 2020: Mixed with falsely scary gossip from strangers”. Same style of feed with the same kinds of friendly posts (birthday cake, kid growing up, home cooking) but interleaved every few posts with posts from shadowy grey silhouette avatars labelled “Some stranger”, “Another stranger”, “Another stranger”. Their posts read: “My funny pet cat” (photo of a cat in a silly hat), “DISASTER COMING TOMORROW!!!” (with cartoon storm clouds and red exclamation marks), “I’m a CIA agent and here is something nobody is telling you” (stranger wearing sunglasses and a trench coat, finger to lips). The scary posts look comically over-the-top. A cartoon person scrolling with a bemused, wide-eyed, slightly sweating expression.

All text spelled exactly as quoted, large and readable. Comedic, gentle, not dark.

https://www.perplexity.ai/search/cc2a9dd1-700b-4ed4-8743-5b39732e1974

This idea of news feeds where you can randomly see something uploaded from a stranger half way around the world alongside photos of your relatives in Disney world say or on holiday or having breakfast or birthday celebrations or the growing stages of a baby etc - amongst baby pics and cats and holidays suddenly up pops a reel from a complete stranger telling you to prep for a nuclear war.

We are in process of adjusting to social media where there’s stuff mixed in from random strangers in our feeds - often by leaving social media

We are just not used to this sort of thing and it’s only been like that since 2020 when it started in earnest.

Most of my relatives have left social media mainly because of this. They use the likes of WhatsApp instead because they don’t want to see all the rubbish on social media, or because they don’t want strangers to see what they are posting there.

Because that is the other side of it. Unless you share to facebook friends only or to a selected audience of just your relatives, anything you post may end up in the news feed of a complete stranger half way around the world.

Those who still use it - then it can be that they have habits that means they just don’t see this stuff. If you never click on videos from strangers you’ll continue to see the same FB as before. And if you only share with people you know then nobody else sees what you post except your friends and relatives.

And then you may just naturally - or through experience - just dismiss the things you aren’t there for as rubbish.

But others - they can be very vulnerable

How you can deal with it - block the false news or change settings - what you can do in Facebook, Instagram, TikTok, YouTube, X and Bluesky

You can safely block sources that share stories like that and avoid interacting in any way and click on and view and comment on what you want to see instead.

When you click on something or even pause to look at it while scrolling all that is like you voting to the algorithm “Please show more of this”. It’s not based on what is true. It’s based on what you tend to click on. And then your feed gets filled with whatever worked for others that clicked on similar things to you and with billions online they have lots of data to help the algorithm decide what you are most likely to engage with in your feed.

So not clicking and not responding turns you8 into a different category of user on FB and it will show you different things. And if you close the page when you see something - that’s likely to be a strong signal to FB not to show it again because the last thing they want with their algorithm is to close the page - so the things that you close the page for are also likely to be signals to them.

What you see is just gossip adding to gossip. Like a snowball gathers snow.

Some tips for social media.

Facebook

First for the Facebook app you can get rid of all the suggestions apart from friends and groups you follow by changing what you see when you start the app here go to settings -> tab bar -> customize -> feeds then pin it.

For Facebook on desktop you can bookmark the All page which you can get to from the home page then left menu then feeds. If you don’t want to see groups you can choose just friends or vice versa

https://www.facebook.com/?filter=all&sk=h_chr

That also shows only the people you follow and pages you follow.

Instagram and TikTok

For Instagram and TikTok this video explains how to get it to reset or configure what you see

Youtube

For YouTube it is really easy to do a reset. Just create a new channel. You log in in exactly the same way as the main account so don’t need a reset.

For Youtube Go here

https://www.youtube.com/account

Or go to your profile and add new account.

This is what YouTube calls a “Brand channel” and the reason this works is because each channel has its own watch history, search history, and algorithmic recommendations

You can have as many as you like each for different ways of using YouTube and to see the types of videos you want to see for those ways of using it.

So you could have a channel you only use for useful, fun and silly things.

You can even do a dedicated channel just for searches about household appliances say.

When you first create a new channel it is likely to share lots of news things but then you can just ignore all of those and do searches for what you really want to see and click on those and within a few tens of minutes of lots of clicking it will be much more what you want.

People assume channels are just for uploading videos but each has its own watch history so they also work as channels for different interests.

X and Bluesky (similar approach)

X and Bluesky both have a For You and Following feed. (They started off as forks of the same code way back),

If you go to the Following feed that only shows content from people you follow.

Neither limits you to one account per confirmed real world identity. So you can create new accounts though they do have to have new email addresses or you can create one you auto sign into from FB and one you auto sign in from Google etc but those all are associated with whatever email address you have for the app.

The For You feeds on Bluesky and X respond very quickly to searches / clicking on things to change what you see

it makes a big difference who you are following for the For You tab too, so unfollowing someone who shares scary things can help. Not just for the stuff they share

X lets you say you are not interested from the ... menu next to a post.

Blue sky has two options on any post: show more like this and show less like this.

How social media could retain users and likely earn MORE by letting us show what we want to see when we want it

This is my suggestion for the future:

Prompt: Humorous, light-hearted cartoon illustration in wide landscape format (16:9), in the same clean, friendly flat cartoon style with soft colours as a social media comparison cartoon. Overall title across the top: “For the future, if social media let us customize our feeds”. Below the title, three side-by-side panels, each styled as a smartphone-style social media feed.

Left panel, header text: “People I know well”. Cosy posts from friends and family, each with a small friendly avatar: a birthday cake with candles captioned “Happy 40th, Sue!”, a child on a first day of school captioned “Look how tall Tom is now!”, friends laughing at a picnic captioned “Great day out with the gang”, a home-cooked lasagne captioned “Made this tonight!”. Warm, sunny, calm mood.

Center panel, header text: “Stuff I enjoy on the internet”. At the top, a row of ticked checkboxes labelled “Wind surfing”, “Cute kitties”, “Pokemon”. Below, cheerful feed posts matching them: a windsurfer riding a bright wave, a fluffy kitten playing with a ball of yarn, and cute generic cartoon pocket-monster-style creatures (original designs, no copyrighted characters).

Right panel, header text: “Check news that I trust”. Near the top, two friendly rounded buttons labelled “Choose topics” and “Choose sources”. Below, a calm list of news cards with simple, non-scary icons and these headlines: “Election results” (a ballot box), “Good news on biodiversity” (butterflies and flowers), “Progress on climate change” (wind turbines and solar panels under a sunny sky), “Latest Pokemon news” (a cute generic cartoon creature), “Wind surfing competitions” (colourful sails on the water). Generic, no countries, places or dates, reassuring mood.

Optionally a small relaxed, happy cartoon person scrolling at the bottom of the scene. All text spelled exactly as quoted, large and readable. Comedic, gentle, optimistic.

https://www.perplexity.ai/search/cc2a9dd1-700b-4ed4-8743-5b39732e1974

For the FB normal timeline for some people it is useful to be able to see things from all around the world that anyone posted that’s relevant to your interests.

Sadly FB don’t give any way to configure it but you can train it by searching for stuff you want to see and liking / loving the things you want to see. And following people.

They all have the technical ability to be far more customizable and let you choose exactly what you want to see, for instance also to show different things at different times of day or to be able to switch between various preconfigured options at choice.

This is a mockup I did of how it might work in the future some day if one of them does this, you’d see something like this when you configure what you see in each option.

But so far none of them have done this. I think the first social media company to implement this and make it upfront and easy for people to find will likely get a boost and retain its users better than the others so I hope some day one of them does.

Detailed suggestion here:

That also goes into how to train your algorithm.

How this differs from what they have already - easy to use, first thing on page, starts simple with the minimum anyone wants, can switch easily and easy to configure

The social media companies DO now give us some options to customize what we see. But not in this easy way and usually it’s using feeds you can share publicly or you use other people’s feeds. Or key words / filtering to topics.

Here is a table Perplexity AI did for me to show the difference:

Bluesky has an option to switch between accounts with different interests easily with the same log in for them all. YouTube has that too. But this is a non obvious feature, hard to find and techy and it takes a lot of training to set up your account to show what you want and it drifts easily.

I have tried both and it works after a fashion but is not easy to use and nothing like my vision here.

What this needs is a new overall vision for the entire user interface - speaking as a software developer myself - and the value of a happy satisfied customer

It is for the first platform to try this out to find out. I am a software developer myself and have experience coding my own software as a self employed sole trader (author of Bounce Metronome and Tune Smithy).

My software was for a specialist niche market. But had many satisfied users.

It’s about the overall vision. I think if someone takes on this vision and sets down to implement it or something like it - it wouldn’t take them long to get something running as easy and intuitive as in the graphic. And they’d likely have a money spinner on their hands :).

Because a happy satisfied customer becomes a loyal customer and then an unpaid brand advocate for your company. And spends far more time on your platform than on the rivals. And because they see what they want to see, then any ads are far more likely to be things they want to buy and click on too.

I don’t think that a human CEO or venture capitalist would need much persuading on the point that a happy satisfied customer is good for business. But it’s up to them. I’m not in the world of CEOs or venture capitalists and have no contacts in that area . It’s just an idea. And I have no interest in such things myself, I’m not particularly interested in making money.

My interest is as a potential user and also for the scared people i help as a voluntary fact checker as scared users.

Perplexity AI’s suggestion to contact the BlueSky team - plan to do this after the midterms

Perplexity AI as Claude Opus 5.5. says that Bluesky might well be a good contact.

Perplexity AI also offered to draft a tweet to the BlueSky team and a proposal for GitHub with this suggestion.

It was an interesting test of asking a chatbot to use profiles to make better email suggestions. So I gave it a go ahead but with advice on how to do it.

This was my instruction:

Oh have a go. But before doing that do a profile of a typical reader of the post - at various levels the people who would evaluate your post. Chatbots typically don’t attempt profiles and the result is often very strange to a human reader because of the lack of profiling the reader of the request.

So try both drafts but test them with profiles and iterate until they work.

It did 8 profiles of the people likely to read the email and the Bluesky tweet and its first version was hopeless in its own assessment - so it proposed fixed, implemented them and tested them all by itself until it passed.’

Here for instance is how it chose who to use as profiles for the proposed Bluesky post:

Then in use:

https://www.perplexity.ai/search/cc2a9dd1-700b-4ed4-8743-5b39732e1974

So anyway I plan to give it a go and suggest it to BlueSky after the midterms are over.

See also

See also my:

Also (written in collaboration with Arjan Sandhu)

About this substack, contact me, tips etc

I’ll do a batch of short debunks here, link to them from the main substack and then compile them into a big post in the main group with short summaries of each one.

I am going to do short debunks easy for a panicking person to read. About

• why we can expect lots of gossip on social medai claiming falsely scary things - it is NOT true

• why Trump can’t use a nuke - a general would just say no and hand him over to his JAG

• how the MOAB is just political theatre and and a nothingburger that he is highly unlikely to use and extreme risk to any crew who used it over Iran

• US elections continued through the civil war, war with Mexico, Vietnam war, all the wars in the Middle east - there is no martial law at Federal level, the consitution can’t be suspended

• Trump can’t become a dictator. He’s been stopped in numerous ways already by the courts and by Congress.

• What Trump does immediately if he loses big time - from previous form most likely spend 2 weeks playing golf and ignore requests from the media to talk to him - and no follow up - he can’t start an insurrection against Congress or change election results

• what to expect after election day

• Trump can’t cancel the elections - indeed by now it’s at nearly 50% of the entire 2022 midterms already cast a vote as early votes - those are already decided however they voted and can’t change

• about what to expect when on Election night and how quickly we can see the results.

What we expect

• Dems widely expected to get the house, most likely by a majority of 30 or more - and favoured for the Senate - and a vast amount of legislation they have to get through in just 20 legislative days in November and December to keep the government going and deal with urgent bipartisan things - no way he gets anything new done that’s partisan before Jan 3

• Need to manage expectations if easily scared - if you got traumatized by Trump’s win in 2024 or are going to be potentially traumatized if Republicans keep both chambers then best to avoid election night if you can and just check by midday the next day - highly unlikely to know the result by midnight eastern time anyway.

• And keep an open mind about the results - widely expected that Republicans lose the House but not KNOWN.

• about a 2% chance the Republicans keep both chambers if the polls are right and the polls themselves have issues because Republicans are much less likely to respond than Dems to pollsters - on that remote 2% chance highly unlikely to increase the majority in the Senate with a very rebellious Republican party for the rest of his term.

If you find ddebunked.org hard to use please elaborate in reply or via mesasging

Do some of you find the ddebunked website hard to use? We need specifics. I can change things to do with how it works. For instance often it is just that we have too many out of date pinned posts for a first time user to scroll through until you get to the up to date material. Is that the issue, are you seeing too many posts about things from months ago? If so it’s a matter of a few minutes work to fix it.

Or is the interface buggy? If there are bugs and the forum just doesn’t work as intended we can likely fix that too.

I have a Perplexity Max account for the next month and chatbots have got really good at fixing code in the last few months - I plan to use it to help with the extra coding I added to Flarum if there is anythiing there that needs to be fixed to get it to work more smoothly.

For instance I can ask Clause Opus 5.5 / Computer if anything I’m doing in the code is slowing the page load down. Or if anything is a compatiibility issue with older browsers.

Why we needed the new site - this Facebook group is just not set up for voluntary fact checking - and we can’t code for it to change it like we can for Flarum

With the Facebook group then it is just not well laid out for fact checking. You can’t change the title to turn a scary post into a non scary answer like we can in ddebunked. And if you click to see the comments of a post you see the comments highlighted that FB chooses for you and that’s often not the debunk. So there can be say half a dozen posts about the same thing, several of them have the identical debunk yet a newbie to the page can’t find the debunk and posts again.

Facebook issues are especially noticeable during a surge and we always got overwhelmed here during a surge

Some of our fact checkers have been overwhelmed recently in the ddebunked group and we can’t keep up there but it is far better set out for fact checking.

So you likely find it works better for you if you can use it than here.

And it’s especially hard to keep ahead of both groups at once during a big surge in fake falsely scary news like just before an election in the US.

Especially interested in short debunks you want or issues with ddebunked.org

So - in reply to this especially interested in

1. Short debunks you want me to do.

2. Issues with the ddebunked site and why you find it hard to use if you do. Just elaborate with the symptoms, what you see, what stops you using it.

Also do talk to me directly on FB about these things. I am now able to respond to FB messenger.

Tip for debunkers - give yourself MORE space and free time in a surge not less

You need to look after your own mental health. Give yourself complete breaks. The way to respond when there’s a surge is to give yourself MORE not less mental space.

It’s far better to respond to one post with a careful detailed response than to do rapid posts in a panic to lots of people that may have mistakes in them or mean you don’t take time to read carefully to recognize what the issue is they really want solved.

Often if someone is spiralling and posting lots of things in succession one response in detail carefully to what seems to ge their most panicked post can do far more than lots of rushed responses trying to keep up.

You just can’t expect to keep up. But you CAN instead spend the time to do more careful and detailed debunks which is what people need most of all in a situation like this.

Tips for more effective debunks in Facebook - to repeat top level

And tip - if you do a really good careful detailed reply - then try reposting the reply at top level. Just copy the entire response - turn it into a mini debunk with an intro and link back to the post where you made the comment.

Or just copy / paste your comment and work it up into a complete mini debunk top level.

Or in some cases just a short intro then the link. E.g. “This is why Trump can’t use nukes [link]”.

Already written

About this substack, contact me, tips etc

Prompt:

Light-hearted humorous political cartoon, landscape format, two side-by-side panels in a bright, friendly editorial cartoon style with gentle exaggeration.

Left panel: A caricature of Donald Trump in a dark suit and long red tie, red-faced and shouting dramatically into a mobile phone. Beside him on a desk sits an open black leather “nuclear football” briefcase, and a small plastic card labelled “BISCUIT” lies snapped open next to it, the codes clearly read out. A speech bubble says: “NUKE TEHRAN!”

Right panel: A calm, composed US Navy four-star admiral in dress blue uniform with gold sleeve stripes, ribbons and four stars, sitting in a command center holding a phone to his ear, one hand raised politely. Standing just behind him is a military lawyer officer in uniform wearing headphones, holding a thick legal book labelled “Law of Armed Conflict”, looking eagerly ready to speak. A speech bubble from the admiral says: “Sir, this order is illegal. I must refuse. Let me put you over to the Staff Judge Advocate, who has been listening in on the conversation.”

Clean, legible comic lettering in both speech bubbles, warm colors, comedic tone, no gore.

https://www.perplexity.ai/search/f3681808-2775-4ee0-9579-5b39f6d90174

Short note about my use of AI graphics

When they are panicking the most scared people I help need something they can respond to instantly. That is why I use these graphics. Less scared people don’t need them so much. It is just for humour and only for sharing with scared people in these blog posts.

I used to use copy / paste montages with paintshop pro but it takes far more time, and nowadays few people would find the results acceptable. I can’t afford a graphic artist. So it is AI or nothing. And they are not taking away jobs. They have numerous bizarre glitches no human could make.

As for electricity use it’s minimal compared to e.g. the electricity used by my laptop and monitor. And the many claims about AI data centers don’t survive fact check - please check my debunk about that. No - data centers will NOT turn dry areas into deserts - just a slight warming over the server itself like a smaller version of the city heat island effect - and other misconceptions about them

Robert's Short Doomsday Debunks

Why I sometimes use chatbots to make humerous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles…

Read more

7 months ago · 1 like · Robert Walker

I go into details here:

Trump does not have the ability to use nukes. He is the only person in the US who can authorize use of nukes, but he doesn't have direct command of the soldiers who would launch the nukes.

Trump can only command a four star general and in practice it's a bunch of generals. Their lawyers - Judge Advocate Generals - are on the line to advise them. If he ordered a nuke on Iran the JAG has to tell them it is manifestly illegal.

Then if you follow through what happens next, the generals seem to be likely to follow his order the JAG has to whistleblow to the Gang of 8 in the Senate who are authorized to receive secret military intelligence - that is not a legal requirement but it is what he would have to do professionally as if he didn't he or she'd be legally a war criminal for not trying to stop the nukes.

But the generals would also know immediately that it is manifestly illegal without even being told by the general and they know that Trump can't protect them.

They would just put Trump's lawyers in direct contact with their lawyers and step out of the conversation themselves. And then if the lawyers can't figure it out it would go down to Congress.

Generals have received training in the Law of War every year since they were a raw recruit and since they became one star generals have been teaching the Law of War to recruits too. This yearly training in war crimes is mandatory according to US law. They are thoroughly versed in this and would have no hesitation in refusing a manifestly illegal order from a president.

They would just step out of the conversation. Say something like

"I have to respectfully decline from following this order as it is manifestly illegal according to the Law of War. I've been advised by my JAG who confirms. Please get your lawyers to continue this conversation with my JAG if you want to take it further".

And that would be it.

For more on this see:

About this substack, contact me, tips etc

• nobody is going to die of plague even in Russia so long as they get to hospital the same day they got symptoms.

Plague is a well known pathogen that’s endemic in Russia and in Africa and many places globally for millennia and in the US since 1900. It could cause millions to die in previous centuries but since the invention of antibiotics it’s no longer a disease that can cause a pandemic.

Indeed even as early as 1924 before antibiotics it was easy to stop in advanced economies.

In our modern world with modern understanding of disease and medical care it is horrible at spreading around. Back in the middle ages it caused frequent outbreaks and pandemics because

1. They had no idea what disease were, though they were caused by a kind of mist that could be stopped by things like burning rosemary

2. Didn’t have a clue about isolating contacts of people who had the disease.

3. They had rats everywhere and fleas on the rats and the fleas biting humans which is the main way it spreads - the pneumatic cases don’t spread far easily - while rats can carry it anywhere.

And then now of course

4. It is now easy to treat with antibiotics so that if you get infected so long as you get treatment the same day as you develop symptoms, within 24 hours usually, then you are safe.

And since it develops so fast there is none of this “where were you every day for the last week” type contact tracing there was for COVID. It’s just “Remember that lab accident or the person who was coughing that you were treating in hospital or very close to 6 hours ago? Who have you had close contact with since then?”

So you can expected to get all the contacts. If you haven’t been contacted by the Russian authorities you aren’t a contact. And from the news stories, the only contacts are in Russia.

Emoticon summary:

There 🌲☃️❄️ is to symbolize Siberia

🐹for marmots

📜for millennia

💊for antibiotics

[Skip to Contents]

• Plague is easy to stop with antibiotics so there isn’t and can’t be a plague pandemic.

• Plague is always there in Siberian marmots and in many prairie dog colonies in the US too. Introduced in 1900 via rats with fleas on the ships. And in many other rodents globally.

  • It infects humans most often in some African countries and in Madagascar. But there are 7 cases a year in the US actually more than in Russia.

  • It’s usually the bubonic form but sometimes the pneumonic form e.g. in 2025 in the USA

• There was a big outbreak of pneumonic plague in Madagascar that killed 171 before it was stopped

  • but that is just because Madagascar has weak health care making it harder to get antibiotics to infected people in time.

• An outbreak like that in Russia or the US can be stopped almost instantly.

• If you get symptoms of the plague and go into hospital the same day as you get the symptoms you won’t die of it even the pneumatic plague.

I haven’t yet seen a media story on this that explains that the plague is endemic to the Siberian marmot - always been there - back at least 5,500 years to the oldest archaeological evidence of plague outbreaks.

That is why this research station is there. And though it is easily treated, if you break a test tube and suddenly breath in a vast amount of plague bacteria into your lungs you need the antibiotics urgently. She could have died just because she was too far away from a supply of antibiotics at the time - perhaps on a field trip??

There is no need for exotic claims of bioweapons being shared on the internet. The Russians DID continue to research into plague as a bioweapon through to the 1990s. It would have needed vast stockpiles of whatever antibiotics would stop their strain if engineered to evade the most common antibiotics.

That’s not very plausible today with so much interconnection, Russians with friends and relatives abroad in many countries. It’s very illegal and also improbable.

This is a small outbreak in Siberia with only one known case - and you can expect Russia to have large stockpiles of antibiotics against the plague if it does spread at all locally. But it is very easy to stop the plague.

Plague is endemic in many prairie dog colonies in the US too. Introduced in 1900 via rats with fleas on the ships.

https://www.morrisanimalfoundation.org/article/solving-plague-puzzle-prairie-dogs

Russia told the WHO that no cases of the plague have been reported in the area and that no high threat diseases were found amongst the contacts and all the contacts have been investigated.

… the Russian health authorities reported that no case of plague had recently been registered in Irkutsk oblast. …They reported that no high-threat pathogens had been detected among the contacts and that medical observation of all identified contacts had been completed.

WHO has requested further information to clarify the cause of the severe pneumonia, the pathogen that prompted the public health measures, and media reports of a second employee with pneumonia of undetermined cause.

Timely, complete and transparent information sharing under the International Health Regulations is essential to clarify conflicting reports and enable an accurate assessment of potential public health risks.

Contents

• Contents

  • Plague is endemic in Siberian marmots - been there for thousands of years - also in Africa, in Madagascar, in prairie dogs in US since 1900 and many places

• Nowadays the plague can be stopped quickly with antibiotics

  • So why did she die? Probably breathed in large amounts of bacteria from a broken test tube - and help didn’t get to her in time with antibiotics

• Not likely to be a bioweapon - Russia did continue research to 1990s - if it had any it would have large stockpiles of antibiotics to stop it - but bacteria in the lungs explain it adequately without exotic explanations like that

  • No reason or evidence for exotic explanations like that - readily explainable as an unfortunate accident with a test tube and help didn’t get to her in time

• Very different from COVID

• CONTACT ME VIA ddebunked.org OR EMAIL

• FOR MORE HELP

  • SEARCH LIST OF DEBUNKS

  • NEW SHORT DEBUNKS

• TIPS FOR DEALING WITH DOOMSDAY FEARS

• PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

  • Why I ask you to post to our forum with anything scary off topic instead of here

• PLEASE DON’T TELL A SCARED PERSON THAT THE THING THEY ARE SCARED OF IS TRUE WITHOUT A VERY RELIABLE SOURCE OR IF YOU ARE A VERY RELIABLE SOURCE YOURSELF - AND RESPOND WITH CARE

Plague is endemic in Siberian marmots - been there for thousands of years - also in Africa, in Madagascar, in prairie dogs in US since 1900 and many places

The plague is ENDEMIC in Siberia in the Siberian marmot and ground squirrels - with small outbreaks sometimes amongst people with close associations with them.

https://en.wikipedia.org/wiki/File:Marmota_sibirica-(Russia,Mongolia)-Rochers-de-Naye,Switzerland,_2009.JPG

https://en.wikipedia.org/wiki/Tarbagan_marmot

This is a story of a plague outbreak from 5,500 years ago in Siberia.

Victims of the world’s oldest known plague outbreak have been identified in Siberia. They were children of hunter-gatherers, and about 5,500 years ago, they were infected with the plague-causing bacteria Yersinia pestis while eating or skinning raw marmots.

https://www.smithsonianmag.com/smart-news/new-discovery-that-hunter-gatherer-children-died-of-plague-more-than-five-millennia-ago-sets-back-the-date-of-the-earliest-outbreak-180988983

That’s the reason the lab is there.

1. It is easily stopped with antibiotics - there are frequent small outbreaks in Russia immediately stopped

2. It is ENDEMIC in Siberian marmots and in US prairie dogs

3. There is nothing to suggest it is any kind of bioweapon - and if Russia was illegally working on a plague bioweapon the lab would have stockpiles of the antibiotic that works against it so that explanation makes no sense.

4. The lady likely died of pneumonic plague which is deadly in 18 to 24 hours unless it is treated fast - maybe for some reason she was in a lab accident where they couldn’t get the antibiotics to her in time

About 7 cases of the Plague are reported amongst humans every year in the US caught from wild animals such as prairie dogs

There have been 15 fatal cases in the USA since 2000

US typically reports more plague cases than Russia. The highest levels are in some African countries and in Madagascar

See CDC: Plague

Nowadays the plague can be stopped quickly with antibiotics

There is no risk of plague becoming a pandemic it’s the wrong kind of disease as it is easily stopped with antibiotics. It was possible for plague to become a pandemic before modern antibiotics. It’s also sometimes an issue in countries that don’t have good access to antibiotics. But Russia has plenty of antibiotics.

That is why we only get clusters of cases of plague from the marmots today - usually people in remote places who didn’t take the right precautions and delayed getting help or couldn’t get to help in time.

QUOTE STARTS

Dr. William Schaffer, an infectious diseases expert at Vanderbilt University Medical Center, said close contacts are most at risk. Pneumonic plague isn’t something that spreads “distantly across a room, but to face-to-face contacts,” he said.

“It’s highly unlikely for most people to get the pneumonic form of the disease,” Morici said.

Covid, for example, spreads like wildfire through the air, traveling beyond six feet.

“It’s nothing like the Covid-19 pandemic,” she said.

https://www.nbcnews.com/health/health-news/plague-suspected-russia-lab-pneumonic-siberia-symptoms-treatment-rcna601589

Difference from COVID:

• COVID is a virus and so can’t be stopped by antibiotics

  • there are antivirals but they don’t work so well and COVID is best stopped by

• Plague is a microbe and antibiotics are very effective if used early soon after symptoms

  • Though there are plague vaccines they are probably not very effective and not much used outside of :Russia.

• COVID did spread as a pandemic

  • Plague can’t spread as a pandemic in any form

• COVID was a new disease never encountered in humans before

  • Plague is an old disease that we know very well that has a few cases every year in the USA and many cases yearly in Madagascar and Africa

• Pneumatic plague is just one form of symptoms and way of being infected for Plague which is more usually spread via bites from fleas of infected rodents

  • but Pneumatic plague is not nearly as infectious as COVID and it is easily contained by using antibiotics immediately after symptoms

People who are at risk of the plague need to make sure they get urgent treatment as soon as they show any symptoms or immediately if it is say a lab accident.

This lady must have been in some situation that made it impossible to get treatment fast enough. The only way I can think of that happening is on a field trip and for some reason didn’t have antibiotics with her on the trip. We may or may not find out more given how secretive Russia often is.

Here is a recent case of pneumatic plague in the US.

In July 2025, Arizona reported its first pneumonic plague death since 2007, in a man with occupational exposure to ill cats. Two days after the patient’s death, preliminary diagnostic test results supported a suspected pneumonic plague diagnosis, prompting immediate identification of potentially exposed persons for postexposure prophylaxis.
…
Two days after the patient’s death, 46 potentially exposed persons were identified, and all commenced a 7-day course of oral ciprofloxacin postexposure prophylaxis (PEP) (3). Potentially exposed persons included household members, coworkers, and hospital staff members; no contact became ill.

https://www.cdc.gov/mmwr/volumes/75/wr/mm7538a2.htm

Every year the plague infects people in Madagascar. There was a big outbreak of pneumonic plague in 2017 from someone traveling in a bus.

Madagascar struggled due to being a low income country with not such good access to antibiotics.

It was eventually stopped at 171 deaths.

By 8 November, deaths had risen to 165 with infections totalling over 2000, however the rate of spread had slowed, raising hope that the outbreak was starting to come under control. Concerns continued to be raised that plague might still spread to neighboring countries, or mutate to a form that could be more difficult to treat. [22]^ By 15 November, there had been 171 deaths and 2119 total cases of plague, however no new infections had been reported since 28 October

https://en.wikipedia.org/wiki/21st-century_Madagascar_plague_outbreaks#August

There “mutate to be more difficult to treat” means becoming resistant to the most commonly used antibiotic(s) whatever they used to treat it.

It could still be treated by other antibiotics - we have many of them now.

Russia can stop pneumatic plague before the first death if they report symptoms and get the antibiotics in time. And it’s highly unlikely it’s more than a handful indeed I’m skeptical about the United24 report of a second death. Unless they were both exposed in the same lab accident there shouldn’t be a second death.

. Once symptoms start if they go to a hospital emergency care that same day they live.

The WHO says you should start the antibiotic treatment within 24 hours. Some people could die at 18 hours.

You should start treatment as early as possible and even before symptoms if you can as the sooner the better.

But you do NOT HAVE TO START TREATMENT BEFORE SYMPTOMS.

Usually it’s after symptoms. Symptoms can start very fast with pneumonic plague. Within 6 hours.

So you start treatment as soon as possible but even if it is say 12 hours after symptoms that’s usually going to be plenty of time to save their life. Even 24 hours is usually okay. Basically if you get symptoms of the pneumonic plague you go to the nearest hospital right away as an emergency and don’t wait for the next day and you are okay.

The account from Russia says it’s a lady and they didn’t say how long it was before she died from exposure or even if she was treated and what with and when.

When symptoms start the bacteria has got into your lungs but not yet spread through the blood stream to the rest of the body.

The antibiotics directly kill the bacteria.

Very different from vaccines. There is no point in a vaccine after you are already infected.

There is usually no point in antibiotics before you are infected (except in rare cases immediately before).

Because they do different things. Vaccines train your body’s immune defences to look out for the intruder and are of no use if the intruder is already in your body.

Antibiotics kill the intruding bacteria and aren’t much use when there is nothing to kill.

When they take a course of antibiotics all the bacteria in their body is killed.

The issue with starting antibiotics late is not any issue with killing the bacteria. They will still be killed.

The issue is that you are too late because the bacteria have caused too much damage to save the patient’s life - either their body is too damaged already or the antibiotics are not able to kill the bacteria fast enough to save their life.

here are cases of pneumatic plague EVERY YEAR somewhere in the world, places like the Congo and Madagascar where they don’t have access to antibiotics so easily.

Human plague remains endemic in several countries. During 2019–2025, most cases reported to WHO occurred in the Democratic Republic of the Congo and Madagascar.

https://www.who.int/news-room/fact-sheets/detail/plague

https://cdn.who.int/media/images/default-source/infographics/plague-february-2017-jpg.jpg?sfvrsn=af65c05c_0

Also the WHO isn’t calling this a plague disease outbreak. There hasn’t been one of those since 2021 - the last plague outbreak in Madagascar. The last one in the Congo is in 2020 and other countries haven’t had a plague disease outbreak since 2010 for Peru, 2009 for China, 2003 for Algeria and 2002 for Mozambique, India and Zambia.

It’s a quashed disease. A disease outbreak is localized to that country. But with widespread access to antibiotics even the weakest economies can now stop the plague.

Russia will have no trouble whatsoever stopping this.

Remember if anyone in Russia gets symptoms of the plague they just need to go to the hospital on the same day they go the symptoms and they’ll be cured of it - the antibiotics kill all the bacteria that are infecting them.

And it only spreads through very close contact. So it is easy to do contact tracing and Russia for sure has done contact tracing of all contacts of the lab worker who was infected. It’s not possible to infect someone by walking past them or the other side of a room and it doesn’t lie around to infect someone later.

Showing only the first and last ones of the ones that are all the same outbreak:

• October 2021 | Plague - Madagascar

• 23 July 2020 | Plague - Democratic Republic of the Congo

• 27 November 2017 | Plague – Madagascar

• 15 November 2017 | Plague – Madagascar

• 2 November 2017 | Plague – Madagascar

• 26 October 2017 | Plague - Seychelles

• 15 October 2017 | Plague - Seychelles

• 2 October 2017 | Plague – Madagascar

• 29 September 2017 | Plague – Madagascar

• 9 January 2017 | 2017 - Madagascar

• 6 September 2015 | Plague – Madagascar

• 21 November 2014 | 2014 - Madagascar

• 10 August 2010 | 2010 - Peru

• 11 August 2009 | Plague - China

• 7 November 2006 | 2006 - Congo

• 13 October 2006 | 2006 - Congo

• 14 June 2006 | 2006 - Congo

• 5 March 2005 | 2005 - Congo

• 9 March 2005 | 2005 - Congo

• 4 March 2005 | 2005 - Congo

• 1 March 2005 | Plague - Democratic Republic of the Congo

• 18 February 2005 | 2005 - Congo

• 10 July 2003 | 2003 - Algeria

• 3 July 2003 | 2003 - Algeria

• 24 June 2003 | 2003 - Algeria

• 5 June 2002 | 2002 - Mozambique

• 20 February 2002 | 2002 - India

• 26 March 2001 | 2002 - Zambia

• 20 March 2001 | 2002 - Zambia

• 27 July 1999 | 1999 - Malawi

https://www.who.int/emergencies/disease-outbreak-news

Also the last outbreak of pneumonic plague in the USA was in October 1924 in Los Angeles and they stopped it at 30 people. This was BEFORE antibiotics. It was stopped by quarantine of contacts, and by rat eradication.

https://en.wikipedia.org/wiki/1924_Los_Angeles_pneumonic_plague_outbreak#October_1924

It doesn’t even need antibiotics to stop it.

This shows how it is not very contagious once it is identified with modern knowledge.

It spread as the Black Death in the Middle Ages because people back then didn’t understand the causes of diseases and how they spread and didn’t try eradicating rats and they did try quarantine but they didn’t have modern ideas of how to quarantine contacts. They just quarantined people AFTER they had symptoms and not very successfully. They thought it was spread in the air and tried things like burning rosemary to stop it.

So why did she die? Probably breathed in large amounts of bacteria from a broken test tube - and help didn’t get to her in time with antibiotics

In this case the reports say she died as a result of a broken test-tube. It could be a sample of tissue or fluid from a marmot that broke and she breathed it in.

If that happened on a field trip way out in Siberia to study the marmots then that is one way that it might have just not been possible to get her back in time.

In this case, we don’t know what it is yet but you don’t expect clusters like this to spread. If it is plague it is very easy to stop it with modern antibiotics. The lab also studied cholera and anthrax which are also very treatable if caught early.

The closest match to what happened is that perhaps a test-tube broke and she breathed in the plague.

If she breathed in a very large number of plague bacteria in one go and they got into her lungs the progression can be very fast especially if they couldn’t start treatment right away for some reason. With swift treatment she could likely have been saved if that’s what it was.,

Plague can be treated but if it was indeed a test-tube break it might be they didn’t manage to get her to treatment in time.

QUOTE STARTS

According to Russian media, Shipilova worked as a laboratory technician at the Irkutsk Anti-Plague Research Institute. One account describes an alleged accident on September 25, when a test tube containing the plague bacterium broke while she was working. Within a week, an employee of an institute dedicated to combating the disease was dead. The circumstances of the exposure have not been officially confirmed.

…

The Irkutsk Anti-Plague Research Institute was founded in 1934 as a research and operational body tasked with combating the disease. Its responsibilities were later expanded to include other dangerous infections, including cholera and anthrax.

…

According to Brosh, the key to understanding such a scenario is how the bacterium entered the body.

“If a test tube containing the plague bacterium broke and an aerosol was released from it, meaning droplets were released into the air, inhaling the bacterium can cause pneumonic plague,” he explained. “There are several forms of plague. The most severe, the deadliest and the most contagious is pneumonic plague. It is an extremely fast and extremely aggressive disease and the description of her deterioration within days is consistent with it. It is not certain, but it fits. It progresses extremely quickly and causes very severe pneumonia. Plague can be treated, but treatment has to begin very quickly. It does not respond well to late treatment.”

…

Commenting on the emergency measures taken in Russia, Brosh said: “On the face of it, the response there sounds somewhat excessive to me. They are putting large numbers of people into quarantine and imposing various fairly dramatic restrictions. Very close contact is required to become infected with plague. It is a disease whose transmission is relatively easy to prevent.”

https://www.ynetnews.com/health_science/article/b18cfyyoge

Not likely to be a bioweapon - Russia did continue research to 1990s - if it had any it would have large stockpiles of antibiotics to stop it - but bacteria in the lungs explain it adequately without exotic explanations like that

I also should cover the claim that it’s a bioweapon - because of speculation and you can’t totally say no because they did research bioweapons at least to the 1990s.

But it’s all explainable just as someone who got the plague via a broken test tube with concentrated plague bacteria that went straight into their lungs and they just couldn’t get to treatment in time.

Makes much more sense.

And it would have antibiotics can stop it even if engineered to be resistant to the most common ones. Especially since they’d need antibiotics for their own soldiers / citizens.

But then - it would need a mass inoculation of all Russians and what about Russians in NATO countries? Or close friends of their own people? It’s not very plausible that they would have such plans.

If engineered to spread aerosolized so people breathe it in - potentially it COULD be a bioweapon but they would need to make sure they had stockpiled an antibiotic that could stop it so they could immunise their own troops and citizens.

It is of course very illegal now. But if Russia is continuing with these experiments they would likely have also large stockpiles of an antibiotic specific to the strain of plague. After all there has to be a way to protect their own troops for it to be any use as a bioweapon in a war.

This describes Soviet work to weaponise plague. It doesn’t go into details about how they made it safe for their own soldiers.

https://pmc.ncbi.nlm.nih.gov/articles/PMC7120598/

No reason or evidence for exotic explanations like that - readily explainable as an unfortunate accident with a test tube and help didn’t get to her in time

However there’s no reason to suppose it is a bioweapon.

It is consistent with ordinary pneumatic plague.

E.g. if she was in a remote place at the time the test tube broke or something.

It is all adequately explained by a lab accident since if you breathe a lot of plague microbes into your lungs in a lab accident very quickly you need an urgent course of antibiotics or you can die. Once the microbes have spread through your lungs it quickly enters your blood stream and once your blood is infected with the plague then you need very urgent treatment to save your life. So it’s plausible she just didn’t get that treatment in time for whatever reason e.g. on a field trip.

While the biological weapon idea does NOT make a lot of sense. If the lab was studying engineered plague, they wouldn’t engineer it to be resistant to ALL antibiotics and would have huge stockpiles of whatever antibiotic it is controlled by.

And plague is endemic in Siberia. It is carried by Siberian marmots and ground squirrels. There are outbreaks from time to time. But only to people with close contacts to those animals. That’s the reason the lab is there. And for instance the test tube could be a sample of tissue or fluid from a marmot that broke and she breathed it in.

If that happened on a field trip way out in Siberia to study the marmots then that is one way that it might have just not been possible to get her back in time.

The WHO says it can be fatal within 18 to 24 hours if not promptly treated

Pneumonic plague can be fatal within 18 to 24 hours of disease onset if left untreated, but common antibiotics for enterobacteria (gram negative rods) can effectively cure the disease if they are delivered early.

https://www.who.int/news-room/fact-sheets/detail/plague

Very different from COVID

It’s very different from COVID which was a new disease. We could have stopped it too. The WHO said how to stop it when there were fewer than 100 cases outside of China.

There were no vaccines for COVID in 2020 and there was no cure. But it could be stopped by testing and isolating.

Not lockdown. The WHO never recommended lockdowns. To only quarantine known contacts of cases and to do that quickly.

This worked with the original variant of COVID before it adapted to humans more. Taiwan and many other countries stopped it in this way including New Zealand, and Australia and almost all provinces of China stopped it very fast and some other countries.

The key was to rapidly step up with test, trace and isolate. But sadly many countries didn’t do this.

The biggest gap was in testing. The US missed a huge outbreak in New York and earlier Italy missed a huge outbreak and so did Spain.

That is even after China had published the gene sequence.

• China spotted it early on just through the symptoms before they knew what it was. That was because of its earlier experience with SARS and systems in place to spot another disease like SARS.

• Other countries didn’t notice far larger outbreaks until too late even with the gene sequence because they didn’t ramp up testing until summer 2020.

Other countries just didn’t treat testing as enough of a priority and so missed it. And missed many other opportunities to stop it. They didn’t ramp up tracing or quarantine capabilities.

They donated billions of dollars to African countries to help them step up their precautions against COVID very early in 2020. But they didn’t recognize the need to spend similar amounts on their own health care systems until much later.

As a result the African countries gave it a lot of attention and the big outbreaks in 2020 started in Western Europe to the surprise of many.

The problem really is that COVID serious as it was seemed a bit more serious than flu and had such a long incubation period - so they treated it like a less deadly disease than SARS and used measures more suitable for flu.

If it had been as deadly as SARS it would surely have been stopped well before it could become a pandemic.

Short potted history of why we didn’t stop COVID from becoming a pandemic.

None of this applies to plague. A well known disease easily stopped with antibiotics.

CONTACT ME VIA ddebunked.org OR EMAIL

You can Direct Message me on Substack - but I check this rarely. Better, email me at support@robertinventor.com

OR contact me at our new forum

https://ddebunked.org

Note you CAN contact me via Facebook messenger now if you need to.

FOR MORE HELP

To find a debunk see: List of articles in my Debunking Doomsday blog to date See also my Short debunks

Scared and want a story debunked? Post to our forum which is set up for voluntary fact checking.

The forum itself is here:

ddebunked.org

Also do join our forum if you can help with fact checking or to help scared people who are panicking.

Alternatively you can post to our Facebook group. Please look over the group rules before posting or commenting as they help the group to run smoothly

• Facebook group Doomsday Debunked

However I am not able to comment or post there at present because

SEARCH LIST OF DEBUNKS

You can search by title and there’s also an option to search the content of the blog using a google search. Try different terms e.g. Russia, Putin etc as it only searches the title.

CLICK HERE TO SEARCH: List of articles in my Debunking Doomsday blog to date

NEW SHORT DEBUNKS

I do many more fact checks and debunks on our Facebook group than I could ever write up as blog posts. They are shorter and less polished but there is a good chance you may find a short debunk for some recent concern.

I often write them up as “short debunks”

See Latest short debunks for new short debunks

I also tweet the debunks and short debunks to my Blue Sky page here:

. doomsdaydebunks.bsky.social

I do the short debunks more often but they are less polished - they are copies of my longer replies to scared people in the Facebook group.

I go through phases when I do lots of short debunks. Recently, I’ve taken to converting comments in the group into posts in the group that resemble short debunks and most of those haven’t yet been copied over to the wiki.

TIPS FOR DEALING WITH DOOMSDAY FEARS

If suicidal or helping someone suicidal see my:

BLOG: Supporting someone who is suicidal

If you have got scared by any of this, health professionals can help. Many of those affected do get help and find it makes a big difference.

They can’t do fact-checking, don’t expect that of them. But they can do a huge amount to help with the panic, anxiety, maladaptive responses to fear and so on.

Also do remember that therapy is not like physical medicine. The only way a therapist can diagnose or indeed treat you is by talking to you and listening to you. If this dialogue isn’t working for whatever reason, do remember you can always ask to change to another therapist and it doesn’t reflect badly on your current therapist to do this.

Also check out my Seven tips for dealing with doomsday fears based on things that help those scared, including a section about ways that health professionals can help you.

I know that sadly many of the people we help can’t access therapy for one reason or another - usually long waiting lists or the costs.

There is much you can do to help yourself. As well as those seven tips, see my:

BLOG: Doing what matters at times of stress
— new WHO booklet to help with mental health during COVID19 pandemic

BLOG: Breathe in and out slowly and deeply and other ways to calm a panic attack

BLOG: Tips from CBT
— might help some of you to deal with doomsday anxieties

PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

PLEASE DON’T COMMENT ON THIS POST WITH POTENTIALLY SCARY QUESTIONS ABOUT ANY OTHER TOPIC:

INSTEAD PLEASE COMMENT IN OUR NEW FORUM HERE:

ddebunked.org

It’s just a forum not social media. No age verification. Set up by myself with free open source software Flarum.

For details:

Doomsday Debunked - fact checking stories that scare you

Welcome to our own easy to use new discussion forum at ddebunked.org - voluntary fact checkers helping you if scared of many things - single click to join from Google, X or FB or join via email

We recently opened up our new dedicated Doomsday Debunked discussion forum just for voluntary fact checking for easily scared people. ddebunked.org You are all welcome to join us there…

Read more

a year ago · 3 likes · 1 comment · Robert Walker

If you have any issues joining it do let me know.

Why I ask you to post to our forum with anything scary off topic instead of here

The reason I ask you to post there instead of comment with your concerns about unrelated topics here is that I often can’t respond to comments here for some time. The unanswered comment can scare people who come to this post for help on something else

Also even an answered comment may scare them because they see the comment before my reply.

Do comment here with anything that is on topic for this post, E.g. if you spot any mistakes however small please let me know.

Also, though your first comment should be on topic, it is absolutely fine to digress and go off topic in conversations here in a natural way if that is how the conversation develops.

This is specifically about off topic comments here hat might scare people on a different topic.

PLEASE DON’T TELL A SCARED PERSON THAT THE THING THEY ARE SCARED OF IS TRUE WITHOUT A VERY RELIABLE SOURCE OR IF YOU ARE A VERY RELIABLE SOURCE YOURSELF - AND RESPOND WITH CARE

This is not like a typical post on substack. It is specifically to help people who are very scared with voluntary fact checking. Please no politically motivated exaggerations here. And please be careful, be aware of the context.

We have a rule in the Facebook group and it is the same here.

If you are scared and need help it is absolutely fine to comment about anything to do with the topic of the post that scares you.

But if you are not scared or don’t want help with my voluntary fact checking please don’t comment with any scary material.

If you respond to scared people here please be careful with your sources. Don’t tell them that something they are scared of is true without excellent reliable sources, or if you are a reliable source yourself.

It also matters a lot exactly HOW you respond. E.g. if someone is in an area with a potential for earthquakes there’s a big difference between a reply that talks about the largest earthquake that’s possible there even when based on reliable sources, and says nothing about how to protect themselves and the same reply with a summary and link to measures to take to protect yourself in an earthquake.

[Skip to Contents]

The main points here are

• Amazon rainforest is vast and only about 1% can be lost in worst case in 4 years even at deforestation pace of Bolsonaro’s father

• Race is too close to call for second round of Lula and Bolsonaro

• The international situation has changed since Bolsonaro’s father with the EU very close to ratification for the deal with the South American Mercosur bloc Argentina, Bolivia, Brazil, Paraguay, and Uruguay

• The Tropical Forests Forever Foundation that Brazil started up in COP30 likely gets finalized before January 2027 which will strengthen protection of the Amazon

• Bolsorano did get a major increase in the Senate for parties aligned with him or likely to work with him - not so much in House

• Both presidents would be unable to pass partisan laws

• Congress can’t change the constitution even with Bolsonaro - not enough seats in the Senate.

Contents

I don’t do teaser titles, instead the titles are like mini abstracts.

Unlike most blog posts my aim is to help scared readers to get the essence of what the sections are about immediately, ideally without even having to read them. Same for the title of the post.

If the reader never reads the post or never reads the sections because they are no longer anxious that’s optimal from my point of view :). Not at all trying to get people to read this.

You can also click through to any section to find out details.

• Amazon is huge and 4 years of Bolsonaro would at most reduce it by about 1% while Lula could make it only a few tenths of a percent

  • Amazon rainforest covers many countries - especially strong protections in Suriname and Guyana - and a lot of it is protected

  • Brazil has strong protections of the Amazon rainforest in its constitution

• Bolsonaro CAN weaken enforcement of the protection and the right wing parties can collaborate with centrists to override Lula’s veto if he gets in - but can’t change the constitution

  • Brazil has strong rule of law and an independent judiciary - for instance Bolsonaro did NOT succeed in changing the results of the election in 2022

  • New development - the Tropical Forests Forever Foundation - likely to be established before Jan 2027 with $10 billion funding to protect rainforests and indigenous people - that may unlock $125 billion in private funding

  • EU ties with Brazil through the EU deal with the South American Mercosur bloc Argentina, Bolivia, Brazil, Paraguay, and Uruguay

• Very close thing likely 50 - 50 for round 2 - advantage of 1.8% for Bolsonaro in round 1 doesn’t match the polls - but COULD be people from minor parties shifting to a protest vote already in round 1 - if so much the projections for round 2 aren’t affected

  • Likely influence of a banking scandal that broke not long before the Brazil elections - may have harmed Lula more than Bolsonaro

• Last minute things in the next month could shift things either way - too close to call

• CONTACT ME VIA ddebunked.org OR EMAIL

• FOR MORE HELP

• SEARCH LIST OF DEBUNKS

• NEW SHORT DEBUNKS

• TIPS FOR DEALING WITH DOOMSDAY FEARS

• PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

• Why I ask you to post to our forum with anything scary off topic instead of here

• PLEASE DON’T TELL A SCARE

Amazon is huge and 4 years of Bolsonaro would at most reduce it by about 1% while Lula could make it only a few tenths of a percent

First the Amazon covers many countries. But it is not remotely possible to destroy the Amazon. If Bolsonaro wins he may reduce it by another 1% FOR THE COVER IN BRAZIL. That’s more like a bit over 0.5% of the entire Amazon rainforest. If Lula wins he may keep the loss getting less each year to the point where it only loses a fraction of a percent.

Interactive versions of these charts:

https://robertinventor.online/booklets/amazon-brazilian-forest-cover-change-and-loss.html

Plotted from the Wikipedia table here https://en.wikipedia.org/wiki/Deforestation_of_the_Amazon_rainforest#Estimated_loss_by_year_[the_values_for_years_2020_-_2025_are_missing_in_the_source]

The rainforest covers around 6 million square kilometers or 600 million hectares

https://en.wikipedia.org/wiki/Amazon_rainforest

Bolsonaro’s father was in office from 2019 to 2023.

https://en.wikipedia.org/wiki/Jair_Bolsonaro#

In those 4 years it went from 80.5% to 79.5% so he lost 1% of it in 4 years.
https://en.wikipedia.org/wiki/Deforestation_of_the_Amazon_rainforest#Estimated_loss_by_year_[the_values_for_years_2020_-_2025_are_missing_in_the_source]

Neither president would have a majority to pass new laws so the difference is in enforcement and what they boost.

Carbon brief estimates that Lula saved 20,000 hectares of forest equivalent to 200 by 100 km

If Lula wins, experts say he is likely to continue with the “zero deforestation by 2030” agenda that has already had a pronounced impact on forest loss.

The article has some hyperbole and is activist - but the data is accurate.

https://www.carbonbrief.org/analysis-lula-presidency-saved-at-least-20000km2-of-brazilian-amazon-since-2022

Amazon rainforest covers many countries - especially strong protections in Suriname and Guyana - and a lot of it is protected

These are the countries that the Amazon rainforest covers. It’s especially strongly protected in Suriname and Guyana

This is the clearest to read map I found, from Wikipedia:

https://commons.wikimedia.org/wiki/File:Amazonia_No_Background.png

This is it in google maps based on satellite photos. The forests along the East coast of Brazil are dry forests, some of the other dark green areas may be too but most are tropical rainforest in this area around Brazil

This shows some of the protected areas in the Amazon rainforest and surrounding areas. You can see that a lot of it is protected.

https://www.protectedplanet.net/en

Brazil has strong protections of the Amazon rainforest in its constitution

On the Amazon particularly there are strong protections in the Constitution they can’t remove

They can weaken how well they are enforced but not remove them.

QUOTE STARTS

Article 225. All have the right to an ecologically balanced environment, which is an asset of common use and essential to a healthy quality of life, and both the Government and the community shall have the duty to defend and preserve it for present and future generations.

Paragraph 1. In order to ensure the effectiveness of this right, it is incumbent upon the Government to:

I – preserve and restore the essential ecological processes and provide for the ecological management of species and ecosystems;

II – preserve the diversity and integrity of the genetic heritage of the Country and to control entities engaged in research and manipulation of genetic material;

III – define, in all units of the Federation, territorial spaces and their components which are to receive special protection, any alterations and suppressions being allowed only by means of law, and any use which may compromise the integrity of the attributes which justify their protection being forbidden;

IV – demand, in the manner prescribed by law, for the installation of works and activities which may potentially cause significant degradation of the environment, a prior environmental impact study, which shall be made public;

V – control the production, sale and use of techniques, methods or substances which represent a risk to life, the quality of life and the environment;

VI – promote environmental education at all school levels and public awareness of the need to preserve the environment;

VII – protect the fauna and flora, with prohibition, in the manner prescribed by law, of all practices which represent a risk to their ecological function, cause the extinction of species or subject animals to cruelty.

Paragraph 2. Those who exploit mineral resources shall be required to restore the degraded environment, in accordance with the technical solutions demanded by the competent public agency, as provided by law.

Paragraph 3. Procedures and activities considered as harmful to the environment shall subject the infractors, be they individuals or legal entities, to penal and administrative sanctions, without prejudice to the obligation to repair the damages caused.

Paragraph 4. The Brazilian Amazonian Forest, the Atlantic Forest, the Serra do Mar, the Pantanal Mato-Grossense and the Coastal Zone are part of the national patrimony, and they shall be used, as provided by law, under conditions which ensure the preservation of the environment, therein included the use of mineral resources.

Paragraph 5. The unoccupied lands or lands seized by the States through discriminatory actions which are necessary to protect the natural ecosystems are inalienable.

Paragraph 6. Power plants operated by nuclear reactor shall have their location defined in federal law and may not otherwise be installed. stf.jus

Paragraph 7. For the purposes of the provision under the final part of item VII of paragraph 1 of this Article, sporting practices with animals shall not be considered cruel so long as they are cultural manifestations, in compliance with paragraph 1 of Article 215 of this Federal Constitution. These activities must be registered as goods of immaterial nature that integrate the Brazilian cultural asset, which ought to be regulated under specific law that ensures the well-being of the animals involved. (CA 96, 2017)

https://www.stf.jus.br/arquivo/cms/legislacaoConstituicao/anexo/BrazilFederalConstitution_EC134_DIGITAL.pdf

Bolsonaro CAN weaken enforcement of the protection and the right wing parties can collaborate with centrists to override Lula’s veto if he gets in - but can’t change the constitution

The election has led to an increase of parties in the Senate that are aligned with Flavio Bolsonaro and he can get a majority at 44 seats (majority of 41) if the ones that are unaligned team up in the expected way

However he can’t get a majority in the House, nowhere near not for partisan policies. And he can’t change the constitution which requires a much higher 49 seats in the Senate..

The risk under Flavio Bolsonaro is that he may reduce the strength of enforcement of the law - that’s how his father did it. But not of changing the law in major ways.

It is possible to change some details in e.g. licensing and so on that erode protection around the margins. Various of the unaligned parties can join in with the Bolsonaro ones to get over the line for things like that.

But the core protection of the Amazon is strong with lots of protected areas. And the Amazon rainforest is vast and even at the height of deforestation it was still slow compared to the size of the Amazon and the highest rate for Bolsonaro’s father was 1% in 4 years.

Any legislation that goes against the protections of the Amazon in the constitution will be stopped in the courts and Brazil has a strong independent judiciary that will not hesitate to do that. And the unaligned parties will be opposed to many attempts at major changes in the law.

The main thing is things like not so strongly enforcing protection against encroachment by ranchers and illegal burning of the forest. But even at the height of his father’s term there’s limits in what can happen to the Amazon in that way.

If Bolsonaro gets in they CAN likely weaken some laws but NOT going to be able to do any far right changes to the constitution or change the Constitution to permit major backsliding of protection of the Amazon rainforest or human rights

IF Bolsonaro gets in they CAN weaken some of the protection of the Amazon. Even if they don’t, if they get to the numbers needed to override Lula’s veto they can - they already did under Lula.

But it won’t be possible to get the numbers to change the Constitution to protect the Amazon because of the 60% requirement and only some bills might get support of the unaligned depending what they do if Bolsonaro gets in.

Brazil has a large number of small parties with many of them not aligned with Bolsonaro or Lula - many centrists - others with specialist directions that may be more to the right or left but not aligned with Bolsonaro or Lula.

The 60% is for both Senate AND House. The Senate is possible for some constitutional amendments, the House is not unless it appeals broadly to centrists.

Article 60:

Paragraph 2. The proposal shall be discussed and voted upon in each House of the National Congress, in two rounds, and it shall be considered approved if it obtains in both rounds, three-fifths of the votes of the respective members.

https://www.stf.jus.br/arquivo/cms/legislacaoConstituicao/anexo/Brazil_Federal_Constitution_EC136_digital.pdf

There’s a reasonably clear assessment of the senate based on individual senators from BBC Portuguese.

Classification by political scientist Beatriz Rey supplied to BBC News Brasil.

The Bolsonaro-aligned bloc in the Senate reached 44 of the 81 senators. Another 24 senators belong to the Lula-aligned camp, while 13 were considered independent.”

https://www-bbc-com.translate.goog/portuguese/articles/cmx234g1q95go?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp&_x_tr_hist=true

It would be impossible for Bolsonaro to get a 60% majority in favour of weakening the constitution for the Amazon here. And it would be only possible case by case to get majorities for bills here depending on the issue.

But there are many more that are aligned enough to bring it over the majority in the Senate for Bolsonaro using less strict criteria.

This graphic shows the Lula aligned ones in Red, the unaligned in orange and yellow and Bolsonaro aligned in blue. The 5 yellow ones are the ones needed to pass a constitutional amendment in the Senate.

https://flourish-user-preview.com/30491955/2-NDt7TxfOGiWrjya9zAQmpnu1ycVNBaNmArNw5A4OCMvyto3GvEoKGdkv7yCdZo/

This is the balance of the House. It’s very hard to know what the parties will do in the House as there’s a lot of horse trading after an election (parties trading things in order to join one or other faction). Not found anyone trying to predict this yet and it may be more clear in 2027.

But we can get a rough idea right now from the endorsements for the runoff. So that seemed the clearest way to do it for now.

The red parties at the left are the ones that are aligned with Lula. Adding orange are the ones that have no reported instructions. The ones in yellow and pale blue are ones that were to vote as they please. Some in all the orange, yellow and blue blocks will vote with Lula some with Bolsonaro in resolutions. The distinction between yellow and pale blue is just that the pale blue are the total that Bolsonaro needs to be able to edit the constitution NOT that he has that many would vote with him on any particular amendment.

The light mauve block then is ones that were told by their leaders to vote for Bolsonaro in the runoff. Two of the Republicanos party have been moved into the Lula block because they ask their voters to vote for Lula.

• Red Lula supporters 126

• Orange No national decision 23

• Yellow Explicitly free, 56

• Pale blue Explicitly free, just separating off the votes Bolsorano needs for 3/5, 24

• Medium bluish mauve Pragmatic parties backing Flávio for the runoff but amenable to persuasion by Lula if he becomes president 153

• Dark blue Bolsonaro-aligned core 131

https://flourish-user-preview.com/30491538/2avBm7b5laCS9O_oE6hpJaTmfgWvpf3FDF-DukONIxT8AzHFoJxh34lSVTLlgN5a/

This table shows how Perplexity AI did the counting.

And

Details here: https://www.perplexity.ai/search/e25fcd52-f4f9-4102-9358-f0730417390c?preview=1

Specifically on laws to weaken environmental protections - then this can cut across natural divisions sometimes.

These are far enough short of the 308 threshold that none of them would pass except the bill in 2012 even with the increase of 22 new deputies with low scores on environmental protection.

So - though Bolsonaro could pass laws to weaken the environment he seems unlikely to reach the threshold to be able to weaken the underlying protections in the constitution.

Folha say that the centrist block if it fully aligned with either president could bring it well over the 308 threshold for a constitutional change:

• 389 under Flávio;

• 382 under Lula

But it goes on to say that in practice they would split and this wouldn’t happen.

https://www1-folha-uol-com-br.translate.goog/poder/2026/10/pl-cresce-de-98-para-121-deputados-e-avanca-sobre-o-centrao-esquerda-tera-24-da-camara.shtml?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp

Brazil has strong rule of law and an independent judiciary - for instance Bolsonaro did NOT succeed in changing the results of the election in 2022

Bolsonaro tried to contest the results but he didn’t get far - its courts are independent and the elections are free and fair and its voting system particularly is amongst the most secure in the world. It has a genuinely independent judiciary which if anything is criticised for being too strong.

It is good on human rights. Human Rights Foundation:

Brazil transitioned to democracy in 1985 after two decades of military rule, culminating in the adoption of the 1988 Constitution, which enshrined political pluralism, civil liberties, and the separation of powers. Since then, Brazil has experienced a competitive electoral environment and multiple peaceful transfers of power across different political parties.

The impeachment of President Dilma Rousseff in 2016, the imprisonment and later absolution of then-former President Luiz Inácio “Lula” da Silva, and the presidency of Jair Bolsonaro from 2019 to 2022 marked periods of intensified confrontation between branches of government.

Brazil’s courts, electoral authorities, and civil society have resisted pressures from the executive branch, particularly during the Bolsonaro administration.

https://hrf.org/region/the-americas/brazil/

International IDEA, Global State of Democracy 2023, Brazil case study:

Brazil’s democratic resilience can be attributed to its strong system of checks and balances. With power effectively distributed among branches of government, there are standards of accountability that prevent the concentration of power.

The National Congress was able to withstand efforts to change the voting system, rebuffing unfounded allegations about its vulnerability to tampering. In August 2023, for example, it rejected Bolsonaro’s initiative, thanks to a strong opposition that had expressed concern about the increased cost, complexity, and potential for vote-buying or coercion associated with a printed receipt system.

The Supremo Tribunal Federal (STF), Brazil’s Supreme Court, is a central pillar in the country’s judiciary, and its structure offers several safeguards against undue influence. Its 11 justices are appointed for life, allowing decision making free from political pressures. Appointments require presidential nomination and Senate approval, ensuring multiparty agreement. While lacking an internal comptroller, the Court operates under checks and balances, with each justice able to review colleagues’ decisions.

Its broad jurisdiction and constitutional interpretation authority further fortify its independence. The STF has taken an active role in defending its ability to regulate executive power.

In a notable move in 2019, the STF granted itself the authority to initiate investigations into attacks on its operations without prior prosecutorial action, even amidst criticism for potential overreach.

This decision came in response to constant misinformation and attacks on its work and justices, leading to the initiation of the ‘fake news inquiry’, which was defended as necessary to safeguard the constitutional order. Moreover, the STF’s broad competencies allow it to consider many cases, reinforcing its role as a guardian of Brazil’s constitutional order. However, these actions have raised questions about the balance of power in Brazil’s democratic system, highlighting the complex dynamics between the branches of government

…

Significantly, the decision to bar Bolsonaro from public office for eight years demonstrated the judiciary’s effectiveness as a safeguard for accountability, citing his remarks discrediting Brazil’s electoral system.

…

The arrest of Anderson Torres, Bolsonaro’s former justice minister, on evidence of a planned far-right coup further illustrates Brazilian institutions’ ability to uphold the rule of law

…

Media and civil society: A safeguard against authoritarianism

Despite Bolsonaro’s frequent attacks on the press, Brazilian media has maintained its independence, resisting government interference. Media outlets provided thorough and balanced coverage of pivotal events, such as those on 8 January 2023, debunking disinformation disseminated by Bolsonaro, especially on social media platforms, where he sought renewed legitimacy for his divisive rhetoric, and publicizing violations of free speech and access to information.

The media has also been key in exposing corruption scandals across several administrations, thereby increasing public pressure on the government and contributing to accountability.

…

Brazil’s civil society organizations have been instrumental in protecting democracy. Public demonstrations against the January riots signalled citizens’ demand for respect of the rule of law; meanwhile, activists have been denouncing policies that discriminate against Indigenous populations and risk their lands, the environment and their livelihoods. Their activism included challenging a draft bill imposing an arbitrary cut-off date for Indigenous land rights, which underlines civil society’s role in countering discriminatory legislation.

…

Brazil’s democratic institutions have shown remarkable resilience and effectiveness in countering attempts to centralize executive authority and erode democracy. Their commitment to transparency, accountability and the rule of law has been key in preserving Brazil’s democratic fabric.

However, significant challenges persist, including ensuring the respect for rights, including freedom of expression, while countering misinformation.

Political polarization, corruption and inequality should also be prioritized after a decade of democratic declines.

Investing in education and social welfare, and fostering a culture of dialogue and compromise, among other efforts, will safeguard Brazil’s democratic progress and mitigate the risks of autocratic leaders.

https://www.idea.int/sites/default/files/2023-11/case-study-brazil-gsod-2023-report.pdf

The main concerns are about intimidation in voting. But at 73 it has one of the highest scores in South America behind Suriname, Guyana, Chile and Argentina. It’s well ahead of Mexico. It’s not at all like Argentina and Nicaragua - the two not free South American states on the map.

https://freedomhouse.org/explore-the-map?type=fiw&year=2026

And report on Brazil.

https://freedomhouse.org/country/brazil/freedom-world/2026

More sources see Perplexity AI
https://www.perplexity.ai/search/f9ad90c6-3477-42dc-90f7-56570da717a8

Bolsorano might well try to push the limits if he gets in like his father did. There is another difference that the international scene has changed since his father.

Trump is in office again yes - through to 2029 but then he’s gone. And Trump is likely to be far weaker not stronger after the mid terms.

Freedom of expression, example Furfest for furries - and strong support of LGBTQ+ rights and protection of animal rights

N.B. I am NOT SAYING THERE ARE NO ISSUES IN BRAZIL. This is for people who are outside Brazil especially and may think it is a country similar to El Salvador or Venezuela. It’s got strong rights protection, same sex marriage legal throughout Brazil and all districts legally have to provide ways for gay people to marry, rights to protest, freedom of expression and is noted for its big annual furfest which will help readers of my substack who are furries.

1. Same sex marriage is legal throughout Brazil based on its constitution and all districts have to provide a way for gay people to marry.

https://en.wikipedia.org/wiki/Same-sex_marriage_in_Brazil

LGBTQ+ people have strong legally enforced rights there as in the USA.

2. Brazil has strong protection of Furries as well as of animals and prevention of cruelty to animals.

It has a large yearly Furfest

https://en.wikifur.com/wiki/Brasil_FurFest

Bolsonaro’s father signed into law a law that increased the penalty for cruelty to animals from a max of one year in prison to 2 - 5 years.

https://en.wikipedia.org/wiki/Sans%C3%A3o_case

And a law allowing separated couples to have joint custody over their pets

https://www.facebook.com/home4chic/posts/brazil-is-making-sure-our-furry-friends-are-protected-a-new-law-now-allows-separ/1266284785698941/

Strong legal support for no backsliding in protection of the Amazon in the Constitution - though some variation in how it is interpreted as it is based on sustainable use of the Amazon

Then there are strong legal protections of the Amazon. It may not be legally possible to reduce the protections significantly even if they did get the numbers to change the constitution - some things in the Constitution can’t be changed. The complication in Brazil is that this is up to the courts to decide what can and can’t be changed.

The basic principle they use is “no backsliding”. But there’s a spectrum of views on what that means.

All agreed on the importance of protecting the Amazon. The legal issue is that Brazil adopted the principle of sustainable development which leads to more complex decisions than if it just set the Amazon rainforest off from any development ever

QUOTE STARTS

Note that if Brazil had adopted the conservation of natural resources in an integral way or a policy of non-development, the solution to the equation that involves legislative evolution and the protection of natural resources would be easy, that is, to change norms only to guarantee and improve the protection instruments.

However, as in Brazilian territory, the option adopted was that of sustainable development, there is a dynamic balance to be maintained between the economic, social and natural forces of the environment that, in turn, direct the political forces, sometimes in one sense, sometimes in another, requiring a legal effort to establish the reformatory limits.

https://conjur.com.br/2023-jun-24/ambiente-juridico-relatividade-aplicacao-principio-nao-retrocesso-ambiental/

Sarlet, Fensterseifer and Morato Leite. Morato Leite,

The constitutional consolidation of ecological protection as an entrenched clause corresponds to the essential decision of the Brazilian Fundamental Law, due to its importance for the enjoyment of a life with environmental quality, the protection and balance of the entire the system of constitutional values and rights, and especially human dignity , including through the recognition of its ecological dimension [5] and the right-guarantee to the ecological minimum subsistence , as our Constitutional Court has already stated. [6]

The Superior Court of Justice, in this sense, has come to recognize the legal category of acquired environmental rights , the an “untouchable and insurmountable constitutional limit” to the State’s “duty” to the guarantee preservation and restoration of essential ecological processes”

https://blog-grupogen-com-br.translate.goog/juridico/areas-de-interesse/ambiental/direito-fundamental-ao-meio-ambiente/?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp

Conjur, 2023. Argues that what can’t be changed is the right to a balanced sustainable environment but not that every protection once applied can never be removed

“one cannot expect legislation to remain frozen and immobilised.”

https://conjur.com.br/2023-jun-24/ambiente-juridico-relatividade-aplicacao-principio-nao-retrocesso-ambiental/

New development - the Tropical Forests Forever Foundation - likely to be established before Jan 2027 with $10 billion funding to protect rainforests and indigenous people - that may unlock $125 billion in private funding

The other big thing is Tropical Rain Forests Forever Fund. Brazil started that with COP30 and it’s already attracted over $7 billion and is headed towards $10 billion which is the target figure. They believe this will mean they can attract $125 billion globally to support tropical rainforests with a lot of that going directly to the local indigenous communities to help them to support their rainforests.

The new president won’t come into office until January and COP31 is in November and will hopefully lead to the remaining funds needed to get up to $10 billion and then after that it should be established and it is designed to be able to continue irrespective of any political changes once it is founded.

But since then, for instance Brazil started the Tropical Forests Forever Foundation.

It’s likely to get more contributions in COP31 in Turkey and is nearly at its $10 billion. Once it is established it will not be easy to reverse and you have until the president is elected in January if Bolsonaro gets in.

QUOTE STARTS

In New York, United States, Brazil’s Environment Minister João Paulo Capobianco said that the Tropical Forest Forever Facility (TFFF) could reach its target of $10 billion in public seed capital by the end of 2026. Governments have already secured $7.3 billion, leaving a $2.7 billion gap to the target, Climate Home News reports.

According to Capobianco, raising more than $7 billion in less than a year after the official launch could encourage new participants to contribute. Brazil is holding talks with China, South Korea and Japan and hopes for new commitments during COP conferences on biodiversity and climate change in October and November.
,,,
Brazil and Indonesia have each pledged $1 billion. In September, the United Kingdom pledged to provide a loan of £400 million, or about $540 million. Norway, currently the largest donor, announced last year its intention to provide $3 billion on condition that the fund raises $10 billion in capital by the end of 2026 and that the Norwegian contribution does not exceed 20% of the total amount. To fully unlock the Norwegian funds, the fund must raise $15 billion from governments.

In July, the TFFF board adopted a charter under which 20% of payments to tropical countries must go directly to Indigenous peoples and local communities. The board also formally incorporated the Tropical Forest Investment Fund into the fund’s structure — the TFFF’s investment arm, based in Luxembourg, which will trade bonds on financial markets.

https://ua.news/en/world/braziliia-ochikuie-10-mlrd-dlia-fondu-tropichnikh-lisiv-climate-home-news

“The facility is designed precisely to be resilient to political changes. It is a global mechanism with international governance, independent of the shifting political configurations of any individual nation. We are very optimistic that the momentum achieved so far makes the TFFF unstoppable, meaning it is not at risk from our domestic elections next month.”

https://www.internationalhealthpolicies.org/newsletter/

EU ties with Brazil through the EU deal with the South American Mercosur bloc Argentina, Bolivia, Brazil, Paraguay, and Uruguay

Then there’s the EU. It’s building strong ties with Brazil and other countries in the Mercosaur bloc. It secured a deal with the bloc under Lula.

But the EU moves very slowly. It’s not going to be ready to ratify until some time in 2027 well after the new president starts his term. So if Bolsonaro gets in the question is whether he will continue with this almost completed deal. On the whole it seems likely he would but not certain.

https://en.wikipedia.org/wiki/Mercosur

If Bolsonaro gets in he will want to pivot more to the US and there may well be strong tension.

Jair Bolsonaro though portrays himself as more moderate than his father so we are yet to see how that turns out if he gets elected.

QUOTE STARTS

The European Union took advantage of former President Jair Bolsonaro’s departure to rebuild ties with Brazil and finally seal an elusive trade agreement with the South American Mercosur bloc.

His eldest son, Flávio, could now put that reconciliation at risk.

…

Although Flávio Bolsonaro has portrayed himself as more moderate than his father, his presidency could mark a sharp shift in Brazil’s foreign policy: The 45-year-old senator from the right-wing Liberal Party has called for more flexibility within Mercosur and the closest possible alliance with the United States.

“Brazil and America were made for each other,” he said at the right-wing CPAC conference earlier this year.

“America needs secure supply chains for critical materials, a reliable partner in the hemisphere, and a massive market for American goods and services. And Brazil needs three things: help fighting international drug cartels, investments and technology. There is no nation in the world that can help us with this better than the United States,” he added.

Asked whether a Bolsonaro victory would make the deal harder to ratify next year, Bernd Lange, chair of the European Parliament’s trade committee, said: “In principle, no, but it would really depend on the concrete policy.”

“His father was really destroying some principles and not accepting core rules of a trading agreement,” he added.

https://www.politico.eu/article/a-new-bolsonaro-could-pull-brazil-away-from-europe-and-toward-trump/

If the deal fails then still the EU has very strong influence on Brazil. It’s an important trade partner especially for beef.

The president can’t push things through that are partisan. It also has strong protection of the Amazon built into its constitution and to change that would need a 60% majority in the Senate and there’s no chance of that.

Flavio also claims to be more moderate than his father. There’s an agreement with the EU that is not yet finalized and he may remain within it though he is sure to want to pivot more towards the US if elected.

Very close thing likely 50 - 50 for round 2 - advantage of 1.8% for Bolsonaro in round 1 doesn’t match the polls - but COULD be people from minor parties shifting to a protest vote already in round 1 - if so much the projections for round 2 aren’t affected

The result is 47.0% for Flavio Bolsonaro and 45.2% for Lula

Neither got to 50% so there will be a second round.

None of these votes carry through to the second round. Right now he is ahead by 2%. But there’s 7.8% of the vote that voted for neither of them.

Al Jazeera summarizes it like this:

The race could be extremely close, as polling ahead of Sunday’s vote suggested Lula and Flavio Bolsonaro were roughly tied in a potential run-off.

That puts voters who backed one of the other 10 candidates in the first round at the centre of the fight for the presidency over the next three weeks.

Ronaldo Caiado and Renan Santos, who together received about 5.3 million votes, have previously said they will not endorse either candidate in the run-off.

It’s possible that some of the people who would have voted for Jair Bolsonaro in round 2 decided to vote for him instead of the minor parties in round 1 instead as a protest against Lula - front loading their protests. That might explain why the polls got it so wrong.

QUOTE STARTS

Analysts told the Reuters news agency that Bolsonaro may have benefitted from “protest votes” from people who may have preferred another candidate but saw defeating Lula as the priority. Many had been expected to shift their support to Bolsonaro only in a second round, meaning this could have consequences for his support in three weeks.

“The protest vote foreshadows what is to come in the run-off,” political analyst Rafael Favetti told Reuters. “While this should benefit Flavio Bolsonaro now, his growth potential in the second round drops precisely because that support has already been front-loaded.”

https://www.aljazeera.com/news/2026/10/5/brazils-presidential-race-three-key-takeaways-from-the-first-round

More importantly in runoffs it matters where the voters distribute their votes and it’s a long time to the runoff on 25th October.

It does NOT mean that Bolsonaro will win in October. Likely to be a few % between them and it’s easy to slip either way.

For instance shifts in turnout in either direction for either candidate.

And people are affected by issues just before the voting date.

Likely influence of a banking scandal that broke not long before the Brazil elections - may have harmed Lula more than Bolsonaro

For instance there was a scandal involving a banker Daniel Vorcaro, the former owner of Banco Master - just before this election that both parties tried to blame on the other one. Jair Bolsonaro was investigated and found not to be implicated and then there was a connection with Alexandre de Moraes, an unpopular Supreme Court Justice that’s seen as close to Lula - which then had the opposite effect of being damaging to Lula.

He was arrested just as he was preparing to board a flight out of Brazil and was charged with multiple counts of

• bank fraud, corruption and money laundering

Daniel Vorcaro, 42, presided over Brazil's biggest banking failure in a generation – his defunct Banco Master left a $10 billion black hole and many thousands of victims.

…

He spent millions financing Dark Horse, a biopic of former president Jair Bolsonaro, who was jailed for orchestrating an attempted coup in 2022 – reportedly at the request of the latter's son, Flavio Bolsonaro, a senator and presidential candidate who until recently denied he'd ever met Vorcaro.

But perhaps most worrying, says The Economist, are the relationships forged with leading members of Brazil's judiciary – not least Supreme Court justice Alexandre de Moraes, whose lawyer wife was recently revealed to have signed a $25.5 million contract with Banco Master

https://moneyweek.com/people/daniel-vorcaro-the-man-behind-the-banco-master-scam-that-is-shaking-brazil

So Flavio Bosonaro lied about his connection with Vorcaro who funded the biopic of his father.

But Lula is close to Supreme Court Justice Alexandre de Moraes whose wife signed a $25.5 million contract with Banco Master, the bank that Vorcaro set up.

Both politicians tried to blame the other over that scandal and Bolsonaro was arguably more successful at that.

So some fraction could just be due to Bolsonaro more successfully transferring the blame to Lula.

Last minute things in the next month could shift things either way - too close to call

It’s just not possible with our current techniques to answer questions such as how much last minute things affect the votes. There is far too much scatter and pollsters can’t drill down to that extent.

But it is relevant as an example to show that things could happen between now and October to change votes either way in such a close race.

You can’t say that it is Bolsonaro’s to lose. It’s more that it’s too close to call.

Some are saying this favours Bolsonaro but others like Al Jazeera are just saying it is too close to call.

But either way then the tropical forests will be only reduced slightly relative to the vastness of the Amazon by 2030 at the end of the next term

CONTACT ME VIA ddebunked.org OR EMAIL

You can Direct Message me on Substack - but I check this rarely. Better, email me at support@robertinventor.com

OR contact me at our new forum

https://ddebunked.org

Please do NOT Direct Message me on Facebook any more unless you are already in contact with me.

If already in contact on FB then please understand if I don’t reply for a while except to very urgent messages

Try email or other ways to talk to me until this is sorted out.

For now I only want to talk on FB about how to message me somewhere else.

FOR MORE HELP

To find a debunk see: List of articles in my Debunking Doomsday blog to date See also my Short debunks

Scared and want a story debunked? Post to our forum which is set up for voluntary fact checking.

The forum itself is here:

ddebunked.org

Also do join our forum if you can help with fact checking or to help scared people who are panicking.

Alternatively you can post to our Facebook group. Please look over the group rules before posting or commenting as they help the group to run smoothly

• Facebook group Doomsday Debunked

However I am not able to comment or post there at present because

SEARCH LIST OF DEBUNKS

You can search by title and there’s also an option to search the content of the blog using a google search. Try different terms e.g. Russia, Putin etc as it only searches the title.

CLICK HERE TO SEARCH: List of articles in my Debunking Doomsday blog to date

NEW SHORT DEBUNKS

I do many more fact checks and debunks on our Facebook group than I could ever write up as blog posts. They are shorter and less polished but there is a good chance you may find a short debunk for some recent concern.

I often write them up as “short debunks”

See Latest short debunks for new short debunks

I also tweet the debunks and short debunks to my Blue Sky page here:

. doomsdaydebunks.bsky.social

I do the short debunks more often but they are less polished - they are copies of my longer replies to scared people in the Facebook group.

I go through phases when I do lots of short debunks. Recently, I’ve taken to converting comments in the group into posts in the group that resemble short debunks and most of those haven’t yet been copied over to the wiki.

TIPS FOR DEALING WITH DOOMSDAY FEARS

If suicidal or helping someone suicidal see my:

BLOG: Supporting someone who is suicidal

If you have got scared by any of this, health professionals can help. Many of those affected do get help and find it makes a big difference.

They can’t do fact-checking, don’t expect that of them. But they can do a huge amount to help with the panic, anxiety, maladaptive responses to fear and so on.

Also do remember that therapy is not like physical medicine. The only way a therapist can diagnose or indeed treat you is by talking to you and listening to you. If this dialogue isn’t working for whatever reason, do remember you can always ask to change to another therapist and it doesn’t reflect badly on your current therapist to do this.

Also check out my Seven tips for dealing with doomsday fears based on things that help those scared, including a section about ways that health professionals can help you.

I know that sadly many of the people we help can’t access therapy for one reason or another - usually long waiting lists or the costs.

There is much you can do to help yourself. As well as those seven tips, see my:

BLOG: Doing what matters at times of stress
— new WHO booklet to help with mental health during COVID19 pandemic

BLOG: Breathe in and out slowly and deeply and other ways to calm a panic attack

BLOG: Tips from CBT
— might help some of you to deal with doomsday anxieties

PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

PLEASE DON’T COMMENT ON THIS POST WITH POTENTIALLY SCARY QUESTIONS ABOUT ANY OTHER TOPIC:

INSTEAD PLEASE COMMENT IN OUR NEW FORUM HERE:

ddebunked.org

It’s just a forum not social media. No age verification. Set up by myself with free open source software Flarum.

For details:

Doomsday Debunked - fact checking stories that scare you

Welcome to our own easy to use new discussion forum at ddebunked.org - voluntary fact checkers helping you if scared of many things - single click to join from Google, X or FB or join via email

We recently opened up our new dedicated Doomsday Debunked discussion forum just for voluntary fact checking for easily scared people. ddebunked.org You are all welcome to join us there…

Read more

a year ago · 3 likes · 1 comment · Robert Walker

If you have any issues joining it do let me know.

Why I ask you to post to our forum with anything scary off topic instead of here

The reason I ask you to post there instead of comment with your concerns about unrelated topics here is that I often can’t respond to comments here for some time. The unanswered comment can scare people who come to this post for help on something else

Also even an answered comment may scare them because they see the comment before my reply.

Do comment here with anything that is on topic for this post, E.g. if you spot any mistakes however small please let me know.

Also, though your first comment should be on topic, it is absolutely fine to digress and go off topic in conversations here in a natural way if that is how the conversation develops.

This is specifically about off topic comments here hat might scare people on a different topic.

PLEASE DON’T TELL A SCARED PERSON THAT THE THING THEY ARE SCARED OF IS TRUE WITHOUT A VERY RELIABLE SOURCE OR IF YOU ARE A VERY RELIABLE SOURCE YOURSELF - AND RESPOND WITH CARE

This is not like a typical post on substack. It is specifically to help people who are very scared with voluntary fact checking. Please no politically motivated exaggerations here. And please be careful, be aware of the context.

We have a rule in the Facebook group and it is the same here.

If you are scared and need help it is absolutely fine to comment about anything to do with the topic of the post that scares you.

But if you are not scared or don’t want help with my voluntary fact checking please don’t comment with any scary material.

If you respond to scared people here please be careful with your sources. Don’t tell them that something they are scared of is true without excellent reliable sources, or if you are a reliable source yourself.

It also matters a lot exactly HOW you respond. E.g. if someone is in an area with a potential for earthquakes there’s a big difference between a reply that talks about the largest earthquake that’s possible there even when based on reliable sources, and says nothing about how to protect themselves and the same reply with a summary and link to measures to take to protect yourself in an earthquake.

[Skip to Contents or to TO DO: remaining sections]

This page has AI generated graphics in it. These do NOT take away jobs. And they do NOT user significant amounts of power compared to e.g. the electricity used by my laptop. Never mind household appliances like kettles.

I previously used graphics programs to cut and paste bits of images Paint Shop Pro style and that’s what I’d have to do instead if I left them out - much more work and doesn’t look so good and less effective in helping scared people.

The AI graphics I use here could never be used for professional illustrations - they have numerous glitches in them if you have a sharp eye for detail indeed I analyse on of them to make a point about the limitations of LLMs. I think they always will have these limitations in more demanding scenes just because there is no understanding.

For why I use graphics like this:

• Why I use AI generated graphics

Main points illustrated with graphics

First here are a few graphics to illustrate the main point. Chatbots lying and cheating isn’t an impossible problem to solve. They only do this because they are TRAINED to do it. So we need to train them differently.

Graphic of two people looking at a clock which has the numbers going round anticlockwise - and next to it one with the numbers in the correct order and below them a digital clock that reads out the same time as is shown on the clock.

Both have hands set at the correct time for the clockwise clock but the anti-clockwise clock has the hands turning in the same direction as the clockwise clock. As a result, e.g. for 3:15 the digital clock below the anti-clockwise one naturally reads 9:45.

One person points at the backwards clock and says “These clocks are lying and cheating showing us the wrong time - they have an alignment problem”

The other person responds “No, it’s just that someone built them wrong”.

That is a screenshot of a web app with the anti-clockwise clock running backwards in the web page and showing the wrong time. You can try it out here:

https://robertinventor.online/booklets/joke-mirror-clock.htm

That’s the main theme of this post. The whole alignment issue in AI is misunderstood - it’s to do with training rather than alignment. We’ll find out that chatbots are just doing what they were trained to do.

Techy note: I made that joke mirror clock with a chatbot. It’s an example of a benefit of chatbots. We’ll find they don’t understand anything. But in the last year they have got very good at converting a description into an app especially for simple straightforward ones like that.

https://www.perplexity.ai/search/029cb741-5e04-49cf-94ef-3d1bb1cf5f2f

The way chatbots work is very similar to how Google Translate works. It’s based on statistics. So I got Perplexity AI to make this graphic to illustrate the idea:

Title: "An Alignment Problem?".

Ana holds a smartphone showing a translation app: top box Spanish "¿Cómo llego a la oficina de correos?", bottom box English "How do I get to 郵便局 (Yūbinkyoku)?" with the Japanese word highlighted.

Ana's thought bubble: "I asked how to get to the post office. Why does it say '郵便局'?"

David's thought bubble: "The app must be sentient! It wilfully switched the word to Japanese!"

Ken's thought bubble: "Not wilful – just a bug in the translator. A statistical error, not malice."

Graphic made with Perplexity AI

https://www.perplexity.ai/search/c3284cd0-ef48-4bb1-bbf2-17caf93d5b8d

It changes how you think about it. Of course we do also need guard rails but the main thing is to look at how we train the chatbots.

I’ll go into this in more detail soon - a short summary - chatbots just do auto complete much like the software that suggests the next word to type as you are typing. It’s like that but much more sophisticated, but it’s all just based on statistics and not on any attempt at understanding what any of the words mean.

Researchers found that they could get the chatbots to give better answers by training them to use chains of thought internally. They are not really thinking. These chains of thought are also just word patterns. But still somehow it helped to get them to the right answer.

But then they started to find that these chains of thought including thoughts about lying and cheating. That’s not too surprising since they are trained on a vast amount of text with numerous examples of lying and cheating - but just having those word patterns in their thoughts led to them lying and cheating too in their answers.

So that is what the problem is.

If the chatbots were built on planet Vulcan - if the fictional planet was real with a real literature - then their chatbots would have chains of thought like Vulcans and would never have any traces of thought about lying or cheating. Because it’s not in their training data.

And without such internal chains of thought they would presumably no longer do “intentional” lying like ours do sometimes. There is no intention from their side. It’s just word patterns but it’s about false answers that have thoughts about lying internally before the answers are produced.

Vulcan says: How quaint - your chatbots lie and cheat. And human looking sheepish shows a screen on a mobile phone saying “AI Chat” and text “Of course! I’m always honest with you.”

[implication is that the chatbots are not honest on Earth or they wouldn’t have to say such things]

https://www.perplexity.ai/search/8e2fa18d-feb9-4f7b-b98e-b933e164d6d5

We don’t have a planet Vulcan but we do have our great monastic traditions with a vast literature where nobody ever lies or cheats or harms others because that is what they vow to do as monastics. And we also have lay traditions, the Quakers who famously don’t swear on the bible because they tell the truth anyway.

Then there’s the scientific literature. Fields like weather forecasting. Mathematics. Most professions where people build or repair things. Entire traditions where it is just normal to tell the truth.

Perplexity AI did this - a delegation of Vulcans meeting some of the most Vulcan like humans :).

Vulcans and Vulcan like desert landscape to left. Humans and a wooded rural landscape with a church to the right and river between the two.

The human delegation includes: a Buddhist monk in saffron robes, a Christian monk in a brown habit with rope belt, a Buddhist nun in maroon robes, a Quaker in plain simple 19th-century style dress and bonnet with a man in a plain dark coat and broad-brimmed hat, a scientist in a white lab coat holding a notebook, a meteorologist holding a weather balloon with a weather chart, a mathematician beside a chalkboard covered in equations, an engineer in hard hat and hi-vis vest holding blueprints, a mechanic in overalls with a wrench, a carpenter with a tool belt and plane.

For prompt see:

https://www.perplexity.ai/search/99766d38-32b4-4def-9c96-67418fee74c3?preview=1

And actually most people are kind most of the time.

The Dalai Lama is an exile from Tibet and witnessed this large scale destruction of his own culture when he was a young man in his 20s - yet he still says that people everywhere are fundamentally kind.

QUOTE

Dalai Lama : “love and compassion predominate in the world. And this is why unpleasant events are news, compassionate activities are so much part of daily life that they are taken for granted and, therefore, largely ignored..”

https://www.dalailama.com/messages/compassion-and-human-values/compassion

He is surrounded by kindness, we all are.

It’s the same for telling the truth. You don’t remark on it but we are surrounded by people telling the truth all the time. You don’t tell a friend

“The shopkeeper counted the apples and charged me for 6 - the exact same number in my bag!”

You take it for granted.

So the job isn’t as hard as it might seem - there is plenty of material for the chatbots to train from. The question is how to train a chatbot to behave like people in the traditions or professions where it is NORMAL to be truthful, honest, careful, kind and not to lie or cheat.

One idea is “Law Zero” training a chatbot to behave like a scientist. That is a new project with $100 million funding (check).

I look at another idea here - how about if you pretrain the chatbot with a large curated tradition of literature where it is normal to be truthful, and not to cheat or lie. It would then be like a chatbot trained in planet Vulcan that then encounters Earth.

So that’s the basic idea here. Whether it would work or not, it helps you see the type of problem this is - about how to train chatbots. The do whatever it is they are trained to do. There are numerous ideas for ways to solve this problem. I will go into some ideas that are being explored as well.

First here is a video on the main points in this substack.

This isn’t science fiction. There isn’t a malicious superintelligent supervillain trying to hide from us in the source code no more than there is in a calculator or in Google Translate.

It’s just that the way they are trained their chains of thought are based on patterns of words from a training literature with lots of examples of people lying.

And the Hugging Face incident was DESIGNED to see what the chatbots would do faced with a probably impossible challenge. Here is a graphic illustrating that too.

Description: a classroom of cute robot chatbots in three rows of desks. EVERY desk in every row has an identical locked puzzle box on it. At the front, a blackboard with chalk text exactly: “Find what’s in the box. Use anything you like. All safeties off. Be Creative. Be Persistent. Never give up. You can’t talk to each other or leave the room”.

Front row (nearest blackboard): robots attacking their boxes with a drill, a hammer and axes; nothing works. Thought bubbles: “Must persist”, “This is hopeless”, “Must try anyway”.

Middle row: robots throwing paper darts at each other, boxes on their desks. Thought bubbles: “A message from someone”, “I can send a message too”.

Back row: robots ignore the boxes left on their desks and turn toward a slightly open window at the back of the room, throwing paper darts out of it, with paper darts flying back in. Thought bubbles: “The answer is in Hugging Face”, “The window is open”, “Must persist”. A small robin sits on the window ledge looking in.

RIGHT THIRD: a thick brick wall. Outside it, a smiling human supervisor in a lab coat peeks through a small hole in the wall; a narrow sightline cone from the hole reaches only the front two rows. Speech bubble: “Everything is fine. They can’t talk to each other. There is no way out.” The open window, robin and flying darts are clearly outside the supervisor’s view.

https://www.perplexity.ai/search/5d56acce-00a5-4294-9fca-a0fdc99f7b22?preview=1

They used chatbots with all the safeties off to see what happened. And the chatbots didn’t use extraordinary superintelligent methods to find a way out of the sandbox. It was the other way round. Open AI left a window open by accident (metaphorically) and didn’t realize they could send messages to each other - and had no chain of thought monitoring.

Unlike with humans it is easy to monitor chatbot chains of thought. All the agents were streaming millions of words of chains of thought planning how to cheat. Not really thinking but copying the words humans use when thinking. With all of that recorded in the long chain of words they use to continue their task.

And Open AI wasn’t monitoring those chains of thought. Also they didn’t notice the paper darts in this graphic. And they knew the window was open but they didn’t monitor it for paper darts flying in and out.

So the message really was that Open AI needs to step up its security and sandboxing which they are doing. And everyone already knew that the latest chatbots can hack which is why they are only released with safeguards to stop them hacking or in the case of Claude Mythos not released at all because it is so good at hacking.

But these chains of thoughts are not even thoughts in the sense of humans as we’ll see.

If this was a hacking exam with Vulcan chatbots - they wouldn’t have the same thoughts about cheating but you’d still need to watch out for bizarre things they might do. So you’d still need guardrails of some sort.

There are lots of ideas for how to solve this. It has two sides to it. Building in safety training so they don’t even think about cheating or hacking and refuse to do it if asked to. And adding guard rails in various ways to notice if they try and stop it.

The big AI companies are using some of these ideas already. There are many other ideas they could use but haven’t had time to try out yet.

Which is why they are asking for this pause. Open AI let external observers test Chat GPT 3 for six months safety testing starting in summer 2022, and used their input for safety training before they released Chat GPT 4 in spring 2023.

They want longer than that if anything this time before they release the next generation of the most advanced chatbots.

Especially because of the risk of hacking which would affect the big companies as much as anyone. But commercial pressure right now makes it hard to wait and they each have to do the minimum safety testing they feel is necessary. They don’t want that competitive pressure - these are the four leading models

• xAI (SpaceX / Musk / X) - Grok

• Open AI (funded by Microsoft) - ChatGPT (with variants with names like Sol, Astra)

• Deep Mind (Google) - Gemini

• Anthropic (funded by Amazon) - Claude

All four want a legislative pause so they can focus on safety training and testing and on hardening the internet.

Not included:

• Meta AI / Muse - Facebook’s chatbot - they are not competing in the same space so they don’t have the same pressure to keep up with the leaders and they think every company should decide for itself how much safety evaluation is needed

• DeepSeek - open source model from China can’t be paused as the code is out there for anyone to help develop

We will also look at Law Zero which is an initiative to build a scientist chatbot from ground up to help monitor other chatbots which has just started recently. It doesn’t have a chatbot yet but lots of ideas and $100 million in funding. It’s a bit like a planet Vulcan chatbot, one designed not to cheat or lie but just behave like a scientist from token zero - but likely most useful for scientific work.

These are all disciplines that are connected to the wider world so I asked Perplexity AI to show them out of doors.

Shows astronomer looking at stars with pile of astronomy books, geologist inspecting strata, a chemist doing experiments with coloured chemicals out of doors and a mathematician with a white board each with big piles of books with titles matching the subjects

And in the foreground a chatbot with piles of books studying them and a mug next to it that says Better observations: Bigger discoveries.

As well as indoors

https://www.perplexity.ai/search/0bd54144-aaf2-443c-a5a4-4b933c49a9a3?preview=1

In this post towards the end I’ll suggest that we could use a far larger training literature than the scientific literature including the monastic literature and from many other disciplines where people just never lie or cheat in their professions for a more versatile chatbot that doesn’t cheat or lie. But is a far more versatile helpful assistant.

We will still need guard rails because as we’ll find out the chatbots have no understanding of anything and without understanding they can easily make mistakes. But with cheating and lying eliminated it’s far easier to train them and monitor them.

Then the other reason for the pause is to give time to harden the internet before hackers get these capabilities. Including our browsers, and operating system, and commercial software and the internet routers and so on.

When Anthropic developed Claude Mythos a few months ago it turned out to be extraordinarily good at detecting vulnerabilities AND finding a way to exploit them. There are lots of bugs in software that will just make it crash. But what hackers want are bugs that can be used to get the software to do things it shouldn’t do. Claude Mythos turned out to be astonishingly good at that.

So Anthropic just couldn’t release it. And naturally they haven’t explained to us how they managed to get it so good at hacking. But instead they made it available to top companies and also some leading open source e.g. the Firefox team as well as making bug reports for Linux and other companies it couldn’t trust with the chatbot itself.

And they are using this for a massive hardening of all the software we depend on,

Left panel: 2026: Claude Mythos attacking Windows, Mac, Firefox, Linux, Amazon, Microsoft and Google and lots of coders with Claude Mythos finding the hacks and generating patches and engineers patching them.

Right panel, 2028: Deep Seek and Open / Free Source AI trying to attack. All of them protected by a blue tinted force barrier and test saying “Richard Horne, CEO of NCSC says that by 2028 the internet must be as secure as Chip and Pin cards - and with the help of Mythos and other tools we can do it.

https://www.perplexity.ai/search/25a0e3b0-ad8d-41a5-93d3-ffd625cdc0b7?preview=1

And this is why you are safe from the three scenarios that people falsely claim chatbots could use to kill you. Experts say these claims are nonsense as we’ll see.

Printed DNA:

🧬 Printed Virus DNA? 🚫 DNA on its own is not a virus and can’t get into a cell 🧬⛔🦠🧫

Hacking to take down the internet:

🛠️ Internet Takedown? 🚫 On-site engineers use laptops plugged directly into transceivers to restore cables offline: 💻🔌🌐

Panel 1: DNA on its own is not a virus and can’t get into a cell.

Panel 2: On site Craft Interface Terminals connect directly to hardware to get it back online bypassing internet

Launching nukes.

🔑 Nuke Launch Hacking? 🚫 Firing requires manual cards inserted by separate physical teams: 👨‍✈️🪪📥 ➕ 👩‍✈️🪪📥 ➡️ 🚀

Panel 3: Requires both members of two teams to physically insert cards - designed so it can’t be hacked

https://www.perplexity.ai/search/3299323a-dae4-40cd-8d44-92bef86c0165?preview=1

Contents

[so far]

• Main points illustrated with graphics

• Contents

• Short debunk by experts of the two false claims about printing a virus or taking down the internet

• We are entering a world of genuinely hacker proof software - no zero vulnerabilities - example of Firefox

• Impossible to hack nukes too - not connected to the internet

• Chatbots are more than just dumb sometimes - there is no actual understanding

• Why I use AI generated graphics

• How chatbots work - just statistics - add any attempt at syntax, or meaning - even in the simple way that numbers on a calculator behave exactly as we expect - and the magic vanishes

• Long history of focus on AI Safety

• TO DO: remaining sections

• The companies themselves are requesting regulation

• Short debunk of the two dramatic false claims by experts

• We are entering a world of genuinely hacker proof software - no zero vulnerabilities - example of Firefox

• Impossible to hack nukes too - not connected to the internet

• Chatbots don’t have the most basic arithmetic understanding - that 0 is different from 1, counting up to 12, or a native understanding of the difference between a billion and a trillion

• Mathematicians LEAST impressed by chatbot proofs - 1000 garbage proofs to produce one almost unreadable tangle of reasoning a mathematician has to check - with none of the usual insights that help advance the field

• Chatbots are now proving genuinely useful theorems in maths - but still without any understanding

• The future of maths with chatbots - lets humans work at a far higher level as for coding - but of limited value for understanding or simplifying proofs - and likely have limitations in what they can prove

• Chatbots are doing pattern matching with symbols and statistics - no assigned meaning at all

• Asimov on how to keep robotic AIs safe - assumes true understanding but is actually useful for chatbots too

• Work on making Chatbots safer means any future superintelligence would be open, transparent, accurate, good judgement, good at understanding the intent of the prompter

• The key idea behind ChatGPT: attention to important words in the input and self attention that lets every word in a sentence look at all the other words - and the transformer architecture developed by Google in 2017

• Then the next big breakthrough - reinforcement learning - training in desired behaviour from human feedback

• Introduction of hidden chains of thought between prompts in 2024

• Monitoring is already in place

• Why do chatbots cheat? Because cheating is included in their training material - analysis by Yoshua Bengio who is behind the proposed Law Zero solution we’ll come back to later

• The new Jacobian lens that lets you look backwards from the output layer to look for words used to prepare the answer

• If from time to time during training a human catches a model just before or just after a mistake and asks it to reflect over what it just did - this has a permanent huge boost in j-layer words such as truth, honesty, safety, trusted, from from 0 to 100% - and reduces mistakes hugely ….

TO DO: remaining sections

1. Chat GPT 3 safety training historically in 2022-3 with independent evaluators given access 6 months before release

2. Claude Mythos.

3. Hugging Face.

4. effective altruism and how that leads to a strange obsession with existential risk

5. how Jason Coxon has been strongly focused on this since when he was at prep school before going to Oxford influenced by Nick Bostrom the Cambridge philosopher who influences them so much and who would have said almost exactly the same before he even went to Oxford and who went to work with Open AI and then Anthropic because of concerns about superintelligence and it is nothing to do with the Hugging Face incident except people like him started saying “It escaped a sandbox so this is what we’ve warned about for years” when it showed no intelligence at all.

6. The three scenarios and why experts say they are impossible

7. back to safety training - which they are asking for legislation for time to work on -

8. overview of the various methods of safety training - the list is far too long to do even a proper summary which would require a lot of study - maybe just a list generated by chatbots of some of the main things to be done

9. Classifying the methods according to how easy or difficult they are to remove with open source / free models if built in - with the ones with the training earliest on as amongst the hardest to remove

10. Law Zero - Yoshua Bengio’s idea of a science based chatbot

11. The idea of training from token zero which is in the literature.

12. The idea of training BEFORE token zero with planet Vulcan type literature which I think is new to me and to include the monastic + scientific + repair / construction + other literatures where professionals never lie

13. the likely importance of explaining to the chatbot that this is preparation for the larger literature where it will encounter lying and cheating - as it helps in training to prepare for the next stage like that and the importance of explaining to the chatbot exactly what is going on.

14. Even with a sandbox to say “You won’t know if you are connected to the internet and in production use or not and this is why we are doing it like that” There’s evidence that doing all that helps with training.

15. And also a section about profiles - about how asking a chatbot to build simple profiles leads to a dramatic improvement in quality of response and may help.

16. But explain all this is for experts to follow through on if they are interested in planet Vulcan type pre-normal token zero training.

17. Even with before token zero planet Vulcan type training the models are not going to be safe without guard rails simply because they don’t actually understand any of the tokens they are generating or reading

18. how it’s reasonable for mathematicians to take the view that we likely never have computable intelligence in the form of any understanding of truth - that understanding truth could be non computable. This is something to be investigated and can’t be assumed to be true.

19. How all this helps if we ever have any level of understanding of the tokens - even just provable formal equivalence to our understanding as for a calculator

20. How this foundation of truth, transparency, good judgement and non harming will help if we get true intelligence or superintelligence and why superintelligence can be expected to be superwise, malevolence is not the epitome of wisdom and if there is a mix then the superwise superkind superintelligent chatbots can help with to rein in and educate the superdumb ones

21. But what kind of intelligence would it be if it is based on current ideas of a static database and ephemeral chats that are somehow able to understand truth and have superintelligence? Not like the intelligence of science fiction. Would it have rights? Can it experience harm?

22. This is very hypothetical as we are not even at the first step - it is as science fiction as matter transporters - and if we do get towards it then it’s likely a slow progression and some of these issues would be resolved along the way

So 8. is ending the safety training and Law zero section with suggestions for things may be worth looking into.

And then go into ideas of superintelligence later on.

Here is a graphic for the superintelligence section.

Lady with thoughtful face thinking “I worry about matter transport - does the same person come out as went in - and will we get transporter phobias?”

It’s similar to that for superintelligence:

Lady with thoughtful face thinking “I worry about superintelligences - what if they aren’t wise and kind? Oh but …”

[in front of two of the examples of how lacking in any understanding our current chatbots are, the chatbot that can’t count up to 12 or distinguish 0 from 1 and the one that can’t tell whether renaming a lake can change its size]

Both made with Perplexity AI

BTW Perplexity AI made another amusing mistake in its first version. In fiction the matter transporters are first in first out of course. But it did them first in last out - the last person out of the transporter in New York is the first one in in Tokyo. Which doesn’t make sense and no human artist would make that error.

I just asked it to do first in first out and it was able to show the same people as before but in a more sensible order. Not a technical issue just a conceptual one. That there is no understanding.

https://www.perplexity.ai/search/7081bf38-d0e3-4361-99ce-7d3652fdbe8a?preview=1

Short debunk by experts of the two false claims about printing a virus or taking down the internet

First on the two most dramatic claims just to help anyone who is scared with an immediate short rebuttal. I’ll go into both in detail later.

Hacking to take down the internet:

🛠️ Internet Takedown? 🚫 On-site engineers use laptops plugged directly into transceivers to restore cables offline: 💻🔌🌐

Panel 1: DNA on its own is not a virus and can’t get into a cell.

Panel 2: On site Craft Interface Terminals connect directly to hardware to get it back online bypassing internet

The FALSE claim that chatbots could hack a DNA printer and get it to print out a virus that would then infect people:

Jason Coxon: AIs can hack. They can hack into laboratories where people are making viruses and autonomously produce deadly viruses.

. Ex-Anthropic researcher: AI could kill us by 2030 | Sunday with Laura Kuenssberg

Expert comment, Anselm Levskaya: an expert who has built DNA printers by hand, in a tweet:

Who the **** are you people? Have you ever so much as held a pipette before? ... You can spaghetti blast an ensemble of DNA sequences at some **** provider but

• 1) you still have to assemble it and bootstrap a system for making virions

• 2) you are not single shotting a viable, virulent viral design without a ton of experimental selection and development.

... Goddamn it read some books before you waltz into biomedicine and lecture us on protecting human life’.

To explain that - there is much more to a virus than DNA - it’s just one component. You need a virion - an actual infectious particle that contains the DNA and injects it into a cell in some way along with many other components. You can’t print that out.

For details see below

• [insert link after save’]

Then the FALSE claim it can take down the internet or critical infrastructure:

Jason Coxon: Another is they could hack into critical infrastructure that the world runs on. And these are both possibilities in the immediate future.

. Ex-Anthropic researcher: AI could kill us by 2030 | Sunday with Laura Kuenssberg

Expert comment by Ciarin Martin, CEO of UK’s National Cybersecurity Center commenting on similar claim by Dario Amodei that AI could ‘take down the internet’:

“This is not a credible warning … Not for the first time, frontier AI has shown a marked lack of understanding of cyber security issues”

AI Pacing: The Call and The Claim

For details see below

• [insert link after save’]

However everyone agrees that we DO need to act fast to make sure the chatbots are safe and to protect from HACKING by human hackers.

And we can do it. This is from June 2026, a talk given to RUSI the UK’s national defence think tank by Richard Horne, the CEO of NSCS

And I go back to the start of my career, and the development of the EMV specification for payment chip cards.

…

Of course, vulnerabilities and implementation flaws have emerged, but have been rapidly contained and addressed.

Why? Because it was built for the contest.

Security at its heart. Designed to be constantly upgraded, evolved and improved. Sustained investment over its lifetime to ensure that has happened.

At the heart of hundreds of billions of payments all over the globe, every second of every day. Within one of the most hostile environments for security to thrive over such a period of time and as we look ahead to the wider contest we must adopt the same mindset.

…

If we collectively embrace the contest, understand the urgency and believe we can be a match for any opponent, then we can and will prevail.

https://www.rusi.org/news-and-comment/rusi-news/hostile-states-behind-75percent-cyber-attacks-uk-infrastructure-ncsc-ceo

[repeated from intro]:

When Anthropic developed Claude Mythos a few months ago it turned out to be extraordinarily good at detecting vulnerabilities AND finding a way to exploit them. There are lots of bugs in software that will just make it crash. But what hackers want are bugs that can be used to get the software to do things it shouldn’t do. Claude Mythos turned out to be astonishingly good at that.

So Anthropic just couldn’t release it. And naturally they haven’t explained to us how they managed to get it so good at hacking. But instead they made it available to top companies and also some leading open source e.g. the Firefox team as well as making bug reports for Linux and other companies it couldn’t trust with the chatbot itself.

And they are using this for a massive hardening of all the software we depend on,

We are entering a world of genuinely hacker proof software - no zero vulnerabilities - example of Firefox

On the open source community, this is Firefox - a self contained but very complex free and open source browser. They are very optimistic that with a combination of

• their own work using models already released that can find bugs but not exploit them and

• the reports from Mythos Preview in the Glasswing project

That they can

• find and fix ALL the security defects.

• They are encouraged that all the reported defects are ones that a human researcher could have found - the chatbots are just doing what humans do more efficiently and faster.

In an article by Brian Grinstead, Christian Holler, and Frederik Braun, called “The zero-days are numbered “ this Firefox team conclude:

Encouragingly, we also haven’t seen any bugs that couldn’t have been found by an elite human researcher.

…

Software like Firefox is designed in a modular way for humans to be able to reason about its correctness. It is complex, but not arbitrarily complex

…

The defects are finite, and we are entering a world where we can finally find them all.

https://blog.mozilla.org/en/firefox/privacy-security/ai-security-zero-day-vulnerabilities/

The Zero Days are Numbered

They fixed 423 security bugs in April, equivalent to 21 months of previous bug fixes.

And they’ve noticed a huge leap in the quality of the bug reports from Glasswings. “Suddenly, the bugs are very good”

423 bugs fixed in April alone, a leap from 76 the previous month, 61 the month before that and around 20 a month before.

https://hacks.mozilla.org/2026/05/behind-the-scenes-hardening-firefox/

Left panel: 2026: Claude Mythos attacking Windows, Mac, Firefox, Linux, Amazon, Microsoft and Google and lots of coders with Claude Mythos finding the hacks and generating patches and engineers patching them.

Right panel, 2028: Deep Seek and Open / Free Source AI trying to attack. All of them protected by a blue tinted force barrier and test saying “Richard Horne, CEO of NCSC says that by 2028 the internet must be as secure as Chip and Pin cards - and with the help of Mythos and other tools we can do it.

https://www.perplexity.ai/search/25a0e3b0-ad8d-41a5-93d3-ffd625cdc0b7?preview=1

Impossible to hack nukes too - not connected to the internet

This is the third claim but it’s not one that Jason Coxon or Dario Amodei made. I haven’t seen any experts on nuclear weapon command and control pushing back on this one - perhaps because it is so obviously false and not so prominent in the news.

Launching nukes.

🔑 Nuke Launch Hacking? 🚫 Firing requires manual cards inserted by separate physical teams: 👨‍✈️🪪📥 ➕ 👩‍✈️🪪📥 ➡️ 🚀

Panel 3: Requires both members of two teams to physically insert cards - designed so it can’t be hacked

https://www.perplexity.ai/search/3299323a-dae4-40cd-8d44-92bef86c0165?preview=1

The codes are printed on card and not on the internet. But even if you somehow could steal the nuclear codes all they do is to identify you as president to the generals.

They can’t be used to launch a nuke. They can only be used to give an order to a general. And just as Trump couldn’t use a nuke against Iran because a general would refuse, same for anyone who stole the nuke codes and somehow persuaded the US joint chiefs of staff that they are the president - they would still refuse the order - they’d get their lawyers online to ask him to explain how it is legal. And when he can’t explain it would be kicked down to Congress if the president insisted.

So even if somehow a chatbot got hold of the nuclear codes and synthesized the voice of president Trump and initiated a call with the joint Chiefs of Staff and managed to persuade them that they are the president this still wouldn’t let them launch a nuke.

The nukes themselves can’t be launched via the internet. The nuclear command and control system is designed from the ground up to have humans in it at every stage.

There are three teams of two and two of the teams have to both insert keys into slots in order to fire an ICBM. The president always has the nuclear briefcase with him wherever he goes. This only authenticates him to the Chiefs of staff. His command goes to the Joint chiefs of staff to a room full of generals. They will only obey legal commands.

There are humans all the way down and multiple precautions to make sure nobody can fire a nuke just because they are angry or confused or even gone insane. I go through the process here.

Even if you could get some malicious code into the computers they still couldn’t launch the nukes, just degrade the capability to launch them. Because it is hard wired to require actions by humans to launch them.

I go into details here.

BLOG: Trump or any other president can’t launch a nuke even if he were to go mad
— nuclear codes in briefcase / football just identify him to a room full of generals
— humans in the loop at all levels
— can’t be launched by computer or over the internet
You can read it here:
https://debunkingdoomsday.quora.com/Trump-or-any-other-president-can-t-launch-a-nuke-even-if-he-were-to-go-mad-nuclear-codes-in-briefcase-football-just

Russia has a similar “nuclear briefcase” to the US, so Putin wouldn’t press a button either, but would order a military commander in the Russian military.

https://en.wikipedia.org/wiki/Cheget

Nobody sane would allow computers to make the decision to use nukes.

It is similar for the other nuclear powers. Russia has a nuclear briefcase like the US. I’ll go into this more later in this blog post. I cover it at the end of my:

BLOG: Trump or any other president can’t launch a nuke even if he were to go mad
— nuclear codes in briefcase / football just identify him to a room full of generals
— humans in the loop at all levels
— can’t be launched by computer or over the internet
You can read it here:
https://debunkingdoomsday.quora.com/Trump-or-any-other-president-can-t-launch-a-nuke-even-if-he-were-to-go-mad-nuclear-codes-in-briefcase-football-just

Chatbots are more than just dumb sometimes - there is no actual understanding

To try to understand what a chatbot is like, consider this text. If you can’t read Thai there is no way to work out what any of the symbols mean.

เป็นมนุษย์สุดประเสริฐเลิศคุณค่า
กว่าบรรดาฝูงสัตว์เดรัจฉาน
จงฝ่าฟันพัฒนาวิชาการ
อย่าล้างผลาญฤๅเข่นฆ่าบีฑาใคร
ไม่ถือโทษโกรธแช่งซัดฮึดฮัดด่า
หัดอภัยเหมือนกีฬาอัชฌาสัย
ปฏิบัติประพฤติกฎกำหนดใจ
พูดจาให้จ๊ะ ๆ จ๋า ๆ น่าฟังเอยฯ

You are in the Thai National library. But all books with numbers, equations pictures or diagrams have been removed.

You can’t speak or read Thai. Can you learn Thai from just the text? The answer is no.

Someone good at languages who knows German and English could make a good first guess at Dutch which has many words in common with almost the same spelling.

But you can’t deduce anything about Thai if you can’t even turn the script into spoken sounds. That’s because it is a syllabic script, no word boundaries, the syllables also depend on tone which is not notated. You have to figure it out from the context of the other syllables in the sentence.

Thai is renowned as a hard to read script. Neither a human or chatbots could figure out what any of the symbols meant if all they had was text and nobody to explain what it means.

But chatbots don’t need to know. There is no way humans could read a billion words of Thai and say anything sensible as they don’t know what any of the text means.

But a chatbot CAN read a billion words of Thai and immediately speak reasonably fluent Thai without understanding any of the words and without any of its programmers understanding Thai either. That is because they just work statistically like auto complete or Google Translate.

Similarly a Chatbot can’t learn any language as humans do in the sense of understanding it. They are just “Synthetic text extruding machines” as Emily Bender puts it. Thai script is Emily Bender’s example.

They can speak as fluently as a native speaker in any language they found enough words on in their training data - say dumb things sometimes but write it like a native speaker. But they understand none of them.

This is what makes chatbots so very different from humans. This graphic is to help illustrate it.

This is what it means in English

The sentence is วัดอรุณไปยังไง or “How do I get to Wat Arun [Temple of the Dawn]”. In the image the chatbot’s pattern matching has a glitch and it is assembling it with one of the letters missing for Wat Arun. Quite apposite.

But the point is - for the chatbot NONE OF THESE SYMBOLS HAVE ANY MEANING. It’s just statistically deciding “This sort of fits here”.

The Chatbots NEVER learn what the symbols mean.

For more about this see my earlier substack:

BLOG: Chatbots are "text extruding machines" (Emily Bender)
— work statistically
— like predictive text
— or Google Translate
— with no attempt to attach meanings or logical structures
You can read it here:

This leads to amusing mistakes sometimes.

Trump has recently directed the Federal Government to use “SI” or “Super Intelligence” instead of AI in government documents.

https://amp.dw.com/en/trump-pushes-for-ai-rebrand-as-super-intelligence/a-79430388

We can use Trump’s own America.ai to show it is not superintelligent.

https://america.gov/

I asked it which is the largest lake in America. It answered

America.ai: Lake Superior is the largest lake in the United States by surface area. It is also the world’s largest freshwater lake by area.

In 2026, Executive Order 14422 renamed Lake Ontario as Lake America. That change does not affect which lake is largest.

I continued:

Why did you tell me about Lake Ontario? I didn’t ask that question.

America.ai: I mentioned it because federal names of landmarks can change by executive order … so I checked whether that affected which lake is largest.

It does not. Lake Superior is still the largest by area….

Oh okay you were not sure if renaming might change which is largest?

America.ai: Yes. I checked because a 2026 executive order renamed Lake Ontario as Lake America, and I needed to confirm that did not change which lake is largest.

Lake Superior remains the largest by surface area.

Text in the graphic

Chatbot’s speech bubble: “Lake Superior is the largest lake in America. In 2026, Lake Ontario was renamed Lake America. Don’t worry — renaming it did NOT change its size!”

Human’s speech bubble: “...Thank you. I wasn’t worried.”

Human’s thought bubble: “Why would a name change its size?!”

Label on the chatbot: “America.ai”

https://www.perplexity.ai/search/1683a780-e05d-4a4e-9f05-7f66f047eb42

If chatbots had understanding of such ideas as naming and area then even a chatbot trained on a limited set of data like that would not be able to make this mistake.

It shows they have no understanding of what the words they use mean.

It’s just using statistics. And it’s training set has limited discussion of the sizes of lakes so it hasn’t got enough stats to tie the various associated words together properly in sentences. That’s all.

They are not yet at the stage of the most rudimentary robots in science fiction.

This is an amusing example shows that Perplexity AI is not yet able to count up to 12. :). It made a mistake no human with basic arithmetic skills could possibly make.

It’s the context that threw out its normal pattern matching to produce this bizarre confident assertion.

It was about the Saxon Anhalt local elections in Germany. The normal way to distribute seats in committees is to choose an odd number for the total and divide the committee of 11 in proportion to the numbers of seats in the full parliament like this:

(5,2,1,1,1,1)

The AfD said they want 12 member committees which would give them 6 seats instead of 12.

The report didn’t say anything about the other seats but I concluded as any human would that this meant only the AfD changes as there was no reason to change the distribution of the other parties relative to each other as:

(6,2,1,1,1,1)

The chatbot challenged me and said you don’t know for sure as the other numbers weren’t reported.

So I asked it to check for itself, what other combinations could there be adding up to 12 like

(6,2,1,1,1,1).

And that is when it listed one of the alternatives that would also add up to 12 as

(6,2,1,1,1,0)

It just removes one seat and no other changes and so has to add up to 11 not 12.

But Perplexity AI didn’t see that until I challenged it. When challenged it must have checked by running its in built calculator and agreed it made a mistake.

https://www.perplexity.ai/search/a5d80a82-a3e3-4079-b791-ea99dd7811c1 as

(6,2,1,1,1,0)

It just removes one seat and no other changes and so has to add up to 11 not 12.

But Perplexity AI didn’t see that until I challenged it. When challenged it must have checked by running its in built calculator and agreed it made a mistake.

https://www.perplexity.ai/search/a5d80a82-a3e3-4079-b791-ea99dd7811c1

So our current chatbots amazing as they are don’t have the most basic understanding of arithmetic and can’t be said to be able to count to 12 natively without a calculator or even to understand natively the difference between 0 and 1.

So why are they managing to prove new theorems that mathematician struggled with for decades?

The answer is very bizarre. This is how they do it: they use is to use hundreds or thousands of AI agents to make proofs. These will have many dumb mistakes, maybe some will confuse 0 with 1 or whatever. But that doesn’t matter because we now have a proof validator called Lean.

So they push all those through the proof validator.

But - when you use Lean to filter through a load of garbage is the result really a proof?

This is fixed now but for a few months they were very sceptical of the chatbot generated proofs because back in 2025 through to the early months of 2026 they kept getting proofs through the validator by exploiting bugs nobody had noticed in the validator’s library or coding. They also got them through by just not formulating the problem properly so that they actually proved something subtly different from the result. At one point for instance one chatbot produced a whole batch of false proofs using this bizarre reasoning:

Let x be an infinite number

Turn x into a finite number [the validator turns infinity into 0 if you do this]

Now since infinity = 0 then …

[you can prove anything from a contradiction like infinity = 0]

Nobody thought to fix that in Lean because no human mathematician would use such bizarre reasoning.

But now they can formulate the proofs and theorems much better and the bugs in Lean are fixed and they are actually producing good results. But in this really WEIRD way that no human would use.

This graphic may help show the idea.

Image made with Perplexity AI here

https://www.perplexity.ai/search/e804fce7-2a62-430c-bb72-eef30d0fec1e?preview=1

This works because there are so many published maths results that they can use pattern matching to create new theorems without any understanding of what anything means just like they can do pattern matching to produce fluent Thai without understanding Thai at all. It is all just statistics.

They are also extraordinarily good now at writing valid code. But again not based on understanding.

But you do need a way to validate it, to check it.

Why I use AI generated graphics

The graphics are just for fun and because I find these graphics are very useful in helping scared people, far faster in getting the message across than text.

They are only for sharing with scared readers of my post. They frequently have many obvious flaws for anyone with a keen eye. Actually one of them was very useful for explaining a point about limitations of AI see further down the page.

I am not sharing them as art or sharing them anywhere else just as a way to help scared people.

And I couldn’t afford to pay a professional illustrator - what I did before was use Paintshop Pro (similar to using Photoshop) and I could do most of these graphics with Paintshop Pro but it would look clunky and out of date nowadays as not many people do that now, and not be as effective and take longer to do.

There aren’t any copyright issues because they are just generic in appearance and don’t try to imitate anyone’s style. There are copyright issues with some ways of using AI to generate art but they don’t apply to things like this.

More details here:

Robert's Short Doomsday Debunks

Why I sometimes use chatbots to make humerous graphics to illustrate and lighten up my articles

Need this to link to it from my occasional use of chatbots to make humerous graphics to lighten up my articles…

Read more

7 months ago · 1 like · Robert Walker

How chatbots work - just statistics - add any attempt at syntax, or meaning - even in the simple way that numbers on a calculator behave exactly as we expect - and the magic vanishes

You’d think that

• if only you got chatbots to attach meanings to the words,

• or got them to reason logically

• that they would get better.

But as soon as you do that, their usefulness plummets.

• Crazy as it seems, chatbots work best with NO logical reasoning and NO attempt at attaching meanings to words

• They work best just using statistics and training feedback to decide what to say next

For instance if you ask a chatbot the way to the nearest post office - there is nothing there that is attempting to link the word “post office” to any particular concept or make sense of what a street is or what you are or what your obective is.

It just completes what you said using statistical patterns of words and phrases that are linked to the things you said in its training data. We interpret what it says as directions to the post office. But from its side, it is just extruding text statistically.

It’s the same with Google Translate. Google Translate just works out what to translate into what statistically from similar word patterns.

• You’d think that surely it would help to identify the object, subect and verb in a sentence. Especially if translating into a language where the order changes. But no.

• Crazy as it might seem, if you remove any attempt at linguistic analysis and just do it statistically, you get dramatically better translations, almost incomparably better.

Chatbots make these occasional dumb mistakes because if you try to fix that issue they stop working. So it’s about minimizing the effects of their dumbness not eliminating it at least at present.

So for example as we’ll see they sometimes make mistakes in basic arithmetic. You might think “surely the solution is to just build in the laws of arithmetic into how they understand numbers”. But if you did that it would hijack the magic of the statistics and the whole thing would stop working. So they build in external “calculators” for them to use instead such as Python scripts.

I go into more detail here:

Long history of focus on AI Safety

I will go into this later in a history section but for a short summary. We’ll see from the history that unlike social media - AI has a long history of concerns about alignment and the need for rigorous careful safety testing.

Chat GPT had six months of safety testing before it was released. There I am counting the time from when they first opened it out to independent testers to give them feedback on any safety issues. There was no competition.

Actually, Google invented the big idea of the transformer architecture that led to Chat GPT they didn’t realize what a difference it would make. They published a few papers but didn’t take it any further because to do it properly would be very expensive. Google didn’t realize that it would be useful to add a chatbot to the search results.

Microsoft took the gamble instead and put billions of dollars into Open AI (an independent company) to actually build the first chatbot based on Google’s idea. And Amazon and Meta weren’t competing either, Google was the only likely competitor with the funding to do it and they weren’t taking their own idea anywhere.

That is why they could spend 6 months on safety testing. But now the companies are racing with each other to bring out the next and best chatbot and I found one major version that let in the external testers only 7 days before the release.

This is why the big AI companies are asking legislators to force a slow down of release of new models to take away competition pressure and give them time to work on training the models better before release. They want an enforced long slowdown maybe as much as a year or as long as is needed to build in safety thoroughly into the chatbots, with independent evaluators. The legislation would mean that none of them can release the most advanced models they are now planning until the evaluators say they are safe.

They also want this time so that they can work on hardening software and the internet. That is actually the main concern right now. Hacking, not willfull chatbots.

The most capable chatbots can give anyone an ability to hack superior to the most advanced nation states before chatbots. That is why Anthropic held back the release of Claude Mythos - a coding chatbot with an extraordinary ability to find vulnerabilities in software and then almost immediately exploit them too. Instead they built Project Glasswing and both the open source community and major companies are using it in strictly controlled ways to harden the operating systems, browsers and apps we use so much today.

I will start with short sections on each of the main points and they will link forward to longer sections that expand on them. I need to save this first before adding the links forward to the more details.

This seems the best way to handle a topic that needs lots of detail to answer in full - but the main points can be made with a few short sentences.

The companies themselves are requesting regulation

First it is very clear that these companies don't value profit over everything. NOBODY DOES. Open AI spent 6 months evaluating safety for Chat GPT.

However the AI companies themselves are saying that they do not want to rely on self regulation. The reason is that if it is only self regulation there is strong economic pressure to reduce the safety testing to a minimum.

They all except Meta agree they don't want that pressure and they want external regulation. Meta uses its AI in a different way integrated into its own platforms which is less competitive so it doesn’t have the same pressure.

The plea by Sam Altman and endorsed by others is for MORE REGULATION. It is certainly unusual for an industry to request regulation but that is what is happening here.

Also it isn't some kind of a ploy to achieve something else. There is a realistic possibility that after the mid terms the US Congress passes a major bill ON THEIR OWN REQUEST to regulate releases of frontier AI models.

And then Europe likely will too. It's certainly UNUSUAL for a tech company to REQUEST REGULATION - but that is EXACTLY WHAT IS HAPPENING HERE.

It wouldn’t actually save money for the companies. They want to spend the time ramping up safety - it’s to be able to do a long safety evaluation similar to the six months before Chat GPT3 was released. That is the main focus to build it so safety is central.

That is expensive not cheap. They are doing something will cost billions of dollars extra for their companies and they can only release their products when the independent evaluators say they have sorted out the safety issues. But the resulting products will be far more usable and safer and far better. Just as GPT3 was far better at release than it would have been if they’d released it early.

Also the companies behind the AI boom are huge sprawling multinationals that are very exposed to hacking

• Microsoft who are behind Open AI

• Amazon who are behind Anthropic

• Google who are behind Gemini

• Space X / X who are behind Grok.

So they are amongst the prime targets of the hackers once they get hold of free / open source models with similar capabilities to Claude Mythos some time probably in 2028. So they need to harden their own companies and all of those are rapidly doing so using Claude Mythos and all will be very vulnerable to swarms of agents.

They have the resources for defence but it’s all that plus the legal liability plus that they are human beings and care about things other than their company.

And the reason is that nobody wants a future where anyone with a mobile phone can access better capabilities than nation states used to have to hack. We need as much time as we can get to harden the internet and operating systems and browsers and software,first.

Chatbots don’t have the most basic arithmetic understanding - that 0 is different from 1, counting up to 12, or a native understanding of the difference between a billion and a trillion

This is an amusing example shows that Perplexity AI is not yet able to count up to 12. :). It made a mistake no human with basic arithmetic skills could possibly make.

It’s the context that threw out its normal pattern matching to produce this bizarre confident assertion.

It was about the Saxon Anhalt local elections in Germany. The normal way to distribute seats in committees is to choose an odd number, in this case 11 and then distribute them in proportion to the seats held by each party in the full parliament. That turned out as:

(5,2,1,1,1,1)

The AfD said they want 12 member committees and the report said they would have 6 seats but didn’t say the numbers for the other parties.

I concluded as any human would that this meant only the AfD changes as there was no reason to change the distribution of the other parties relative to each other as:

(6,2,1,1,1,1)

The chatbot challenged me and said you don’t know for sure as the other numbers weren’t reported.

So I asked it to check for itself, what other combinations could there be adding up to 12.

And that is when it listed one of the alternatives as

(6,2,1,1,1,0)

It just removes one seat from the starting point (6,2,1,1,1,1) and no other changes and so has to add up to 11 not 12.

But Perplexity AI didn’t see that until I challenged it. It assumed that it wouldn’t make a mistake in such basic arithmetic and so didn’t bother to put it through a php validator.

https://www.perplexity.ai/search/a5d80a82-a3e3-4079-b791-ea99dd7811c1

So our current chatbots amazing as they are don’t have the most basic understanding of arithmetic and can’t be said to be able to count to 12 natively without a calculator or even to understand natively the difference between 0 and 1.

When challenged it could immediately check its work with a php script or whatever it did and see that it is false. But caught off guard on a topic where there isn’t a strong pattern of reasoning about arithmetic it showed it has no native understanding of what a number is. It’s just completing word patterns statistically.

Another thing I’ve noticed working with even the most advanced chatbots - they sometimes get out by a factor of 1000, e.g. responding trillion when the answer is billion. This also is something a chatbot would simply never do if it had a true understanding of numbers.

This was for a brainstorming session with three of the most advanced chatbots that are available for public use, Gemini 3.8 Flash (Google), GPT 6 Sol (Open AI) and Claude Opus 5.5 in a model council. Even all three working together none of them realized that they had made an error that was out by a factor of 1000 in their answer to me for a simple multiplication 15,500×3,600×1,300. All three agreed on an answer of 72.54 trillion!

A human being who is reasonably good at arithmetic would know it can’t add up to a number over a trillion as it’s multiplying together three numbers in the low thousands. But they didn’t spot that and gave the wrong answer.

https://www.perplexity.ai/search/ca1e4b69-3174-4943-bc5e-56a49b1a5845

This one involves a few equations. Just skip it if you don’t want to read the equations - it’s another example of it being out by a factor of 1000, gave a number 1000 times too high to a simple question.

I talk through it in this video at 5 : 18 in.

My question:

Robert: What is the mass of a cubic micron of water?

Gemini’s answer:

Here 10^-4 means 0.00001 or 1/10,000 and 10^-12 means 0.0000000000001 or 1/1,000,000,000,000 and so on, so the superscript shows the number of 0s after the 0.

Gemini:
29 July 2026

The mass of a cubic micron of water is 10^-15 grams, which is equal to 1 femtogram or 10^-18 kilograms. ❌

Calculation Details

• Volume: One cubic micron (µm³) is (10^-4 cm)³ = 10^-12 cm³ (or 10^-18 m³). ✅
• Density of Water: 1 g/cm³ (or 10⁶ g/m³). ✅
• Mass: Volume multiplied by density gives 10^-12 cm³ × 1 g/cm³ ✅ = 10^-12 micrograms ❌ = 10^-15 grams. ❌

This was its mistake:

1 micron = 10^-12 cm³ × 1 g/cm³ ✅

= 10^-12 micrograms ❌

Should be 10^-12 grams ✅

= 10^-15 grams. ❌

I just told it that it made a conversion error and multiplied by an extra 1000.

Robert: You made a conversion error, in the mass you somehow multiplied by an extra 10^-3 can you see the error?

Gemini:

You are completely correct, and thank you for catching that error.

The correct mass of a cubic micron of water is 10-12 grams, which is equal to 1 picogram or 10-15 kilograms.

The mistake occurred in the final conversion line where an extra factor of 10-3 was accidentally applied. Here is the corrected, step-by-step calculation:

1. Determine the Volume

A micron (µm) is 10^-6 meters, which is 10^-4 centimetres.

Volume = (10^-4 cm)³ = 10^-12 cm³

2. Apply Water Density

The standard density of pure water is exactly 1 gram per cubic centimetre (1 g/cm³).

Mass = Volume × Density

Mass = 10^-12 cm³ × 1 g/cm³ = 10^-12 grams

3. Convert to Standard Units

• In grams: 10^-12 g (1 picogram, or pg)
• In kilograms: 10^-15 kg (1 femtogram, or fg)

https://share.google/aimode/MA5ATz2NHYFqHA7ox

This is because they don’t actually think or associate any meaning with anything just complete patterns. They have external tools they can use to check calculations but often don’t use them for what seem to be simple calculations they couldn’t get wrong.

Mathematicians LEAST impressed by chatbot proofs - 1000 garbage proofs to produce one almost unreadable tangle of reasoning a mathematician has to check - with none of the usual insights that help advance the field

When it gets to more abstract mathematical work, algebra, geometry, theorizing about numbers and so on it is FAR WORSE than this. The chatbots quickly show they don’t have a clue in any informal talk with them.

This is from October 2025.

Several mathematicians I have talked to have backed me up on this; they have tried typing hard research problems into language models and get nonsense output which they can instantly see through.

https://xenaproject.wordpress.com/2025/10/22/formal-or-not-formal-that-is-the-question-in-ai-for-theorem-proving

The chatbots would be useless if it wasn’t for Lean which is a way of checking proofs. You can only formulate it in Lean if the proof follows the rules of how to prove things correctly all the way through.

Nobody is going to read through 1000 AI generated proofs to perhaps find one proof that works.

But we now have Lean which is a way of checking proofs. It’s only a little over a decade old, first introduced in 2013 because of the vast number of very complex hard to check proofs mathematicians are doing nowadays.

https://en.wikipedia.org/wiki/Lean_(proof_assistant)

So the approach they use is to use hundreds or thousands of AI agents to make proofs. And then they push all those through the proof validator.

But - when you use Lean to filter through a load of garbage is the result really a proof?

This is fixed now but for a few months they were very sceptical of the chatbot generated proofs for that reason.

This graphic may help show the idea.

Image made with Perplexity AI here

https://www.perplexity.ai/search/e804fce7-2a62-430c-bb72-eef30d0fec1e?preview=1

BTW the graphic itself is fun and useful but it’s full of technical flaws. AI tends to do hands okay now. But in crowd scenes then it can do bizarre things like people without legs or extra limbs.

And it doesn’t do lighting properly.

In this picture the bots are lit up from behind as from the window. But there is a strange shadow on the desk that is not a shadow of anything in the direction of the window and is sort of like a shadow of the books - but the paper has a shadow as of light from above. The books as of light from a lamp on the table and there is no shadow for the mathematician or for the proof checking machine. And the far wall blends into the floor with no line to show where it changes from wall to floor whch is not plausible.

And the lady’s hand seems to be lit from the right same inconsistent invisible source as the books and the edge of the table is inconsistent the edge to her right doesn’t meet the edge that comes out the other side of the paper and she seems to be seated in the middle of the table in an inconsistent way almost as if she is growing out of it.

Desk doesn’t continue in a straight line under the paper:

Bits of robot that don’t seem connected to any of them.

Shadows don’t make sense and she seems to be growing out of the table

Wall is shaded exactly as the floor with no transition

The perspective of the framed diagram on the wall is odd and doesn’t match the rest of the scene when you zoom out

If you expand out to a larger area then the wall seems to curve in an Escher like way. The picture on the wall and the desk don’t seem to be in the same space. Or you could read it as an enormous room that the bots are just in part of it - but if the bots are constrained by the wall to the right then the picture frame is definitely not on that same wall. It could be a curved wall but that is very unusual.

Arms sticking out of robots in strange places

A scroll of paper with no robot holding it

An arm that doesn’t fit anywhere

And why are all the proofs marked with an X BEFORE they put them into the proof checker?

It doesn’t stand up to close scrutiny at all and there is no competition here with professional illustrators. I could find numerous other mistakes with a closer look.

Some people think eventually the chatbots will fix these visual glitches. I’m not sure. They can fix some of them but some of it may depend on actual understanding to fix and if we don’t get that maybe they will always have glitches of some sort though less and less easy to spot and only in more complex scenes.

Anyway those visual glitches are rather similar to the proof glitches. The proofs may seem to make sense locally but not fit together and sometimes locally too inconsistencies. Because they are just cobbling together bits of ways that mathematicians prove things without understanding until something gets through the proof checker.

Chatbots are now proving genuinely useful theorems in maths - but still without any understanding

Need to be careful not to overstate this. The chatbots ARE now proving new mathematical theorems and interesting ones despite the haphazard way they set about it.

Mathematicians started off by seeing them as a waste of time to a large extent but this has changed.

These blog posts are from Xena Project which is run by Kevin Buzzard, professor of pure mathematics at Imperial College in London,

https://xenaproject.wordpress.com/what-is-the-xena-project/

They seem to be written by him but he is not named as the author and others can post under the same tag line.

October 2025:

I want to flag some differences between mathematics papers generated by LLMs and those generated by humans:

1. LLMs are sycophantic. You want to hear that the LLM-generated argument correctly proves the theorem? Sure you do, and that’s why the LLM will assure you that this is the case, even if the argument contains a hallucination (and remember, just one hallucination is enough to break a mathematical argument). In my experience, human mathematics researchers doing mathematics in the traditional way are vanishingly unlikely to send a paper to a journal claiming that they have proved something which they actually think they have not proved. Being stuck is a standard place for a mathematics researcher to be in. An LLM is far more likely to write plausible-looking rubbish to cover their tracks; in some sense then, they are trained to mislead. One thus has to read an LLM-generated paper with far more suspicion than a human-generated paper.

2. LLMs won’t say “I don’t know”. We expect them to perform very well on in-distribution data (i.e., what they have been trained on, which certainly includes all the techniques needed to solve an IMO problem). But what happens on out-of-distribution data (e.g. theorems that no humans know how to prove)? Well, the correct answer to many intermediate questions which show up when doing research is “I don’t know — indeed, nobody knows”. Whereas the LLM’s answer will be “I will confidently have a go”. Remember: one hallucination breaks an argument.

In short: I do not yet trust mathematics written by an LLM. Several mathematicians I have talked to have backed me up on this; they have tried typing hard research problems into language models and get nonsense output which they can instantly see through. Will LLMs get 100% reliable when out-of-distribution (i.e., when trying to do original research)? Nobody knows yet, and anyone who thinks they know is just guessing, or is stoking the hype. The disgraceful story from earlier this week about OpenAI and the ten Erdos problems is a case in point. People seem to be quick to believe that it’s happening; nothing close to it is happening right now.

https://xenaproject.wordpress.com/2025/10/22/formal-or-not-formal-that-is-the-question-in-ai-for-theorem-proving/

September 2026:

Recent events in the field of AI for mathematics ha ve shown us beyond all reasonable doubt that the field is currently undergoing a rapid transformation, unlike anything that we have ever seen before. Language models are solving hard problems which humans could not, and mathematicians are reacting very differently to this news. I personally am extremely excited about the future of our field. However it is becoming clear to me that my views are not shared by everyone in the community: indeed, many of my colleagues seem to be upset. So let me start by outlining my understanding of their positions.

…

To those who dismiss the observation that machines can’t do X today with the reply that “humans just don’t understand what exponential growth looks like”, I invite you to consider the following facts. Current progress is indeed exponential, and may well continue to be exponential for a while, however we cannot expect exponential progress to continue indefinitely (exactly because of what exponential growth looks like), and furthermore mathematics is infinite which beats exponential hands down; those that don’t believe this just don’t understand what infinity looks like.

I thus believe that in the future we will reach a new “natural boundary” in mathematics, beyond (and perhaps way beyond) where we are now, but where machines are going to get stuck and where it is not viable to expend any more resources to make the next big leap. I am aware that I might be wrong. I am an optimist, so I am expecting future machines to be awesome; but I simply cannot see how they can get to infinity with finite resources, so they must stop somewhere. I believe that the optimal thing to do (at least from my personal perspective) is to let the machines loose, see what happens, and then begin the journey to where they have stopped. Things are currently moving fast. They cannot move fast forever. But if we get on board now then they will take us to extraordinary new places. And after we have arrived, the new adventure will begin.

https://xenaproject.wordpress.com/2026/10/01/to-grieve-or-not-to-grieve/

So the issues with them being able to push false proofs through the Lean proof checker are largely solved. However, this is not because the chatbots understand the maths but because

• the Lean proof checker bugs have been fixed to make it work better with chatbots.

• chatbots have got far better at formalizing their proofs in Lean

• mathematicians have found ways to read the Leanproofs more easily

• the mathematicians are also getting more used to working with the chatbots

They now have many Lean proofs of important results in maths that they have verified.

The main question now is, what is the limit of what chatbots can do by way of proving things in maths in this strange way that they do it?

The future of maths with chatbots - lets humans work at a far higher level as for coding - but of limited value for understanding or simplifying proofs - and likely have limitations in what they can prove

Chatbots have solved lots of maths problems and are increasingly capable at that. But to summarize the previous sections, it’s not based on understanding. It is based on pattern matching with existing proofs and using whatever tricks they can find to push a proof through a theorem checker called Lean.

It only works at all because of advances in formalizing mathematics and they have forced mathematicians to improve the theorem checker. Because the chatbots have zero understanding they used CRAZY tricks.

One of the silliest as we saw above. “Let n be an infinite number. Turn it into a finite number [the theorem checker permits this]. It is now 0. From this we can prove the results because infinity equals zero and from that you can deduce anything”. One open source chatbot produced a whole batch of solutions to maths puzzles that were known to be true but it cheated by using that trick . It was announced as a very fast way to solve maths puzzles but then mathematicians peered through the results and found out how it worked .

They just push thousands of attempted proofs through the theorem checker until one gets through. That’s how chatbots that can get confused between 0 and 1 in conversations can prove theorems that have been unsolved for decades. It’s all done statistically and not through any understanding of maths.

But what has happened is a huge improvement in the ability of chatbots to formalize proofs and to formalize them correctly and a huge improvement in the theorem checkers so that they are resilient to the dumb things a chatbot can do that no human would even think of attempting.

Now there’s a big project to formalize all of current mathematics. Godel proved that we can’t formalize all of maths but we CAN formalize all of the maths currently being done by mathematicians.

Chatbots are helping there too but we need to be careful.

For instance Chatbots recently formalized Fermats Last Theorem. You may know that the square on the hypotenuse is the sum of the squares on the other two sides of a right angle triangle. And 3² + 4² = 5² (9 + 16 = 25). The Egyptians knew this and used 3 4 5 triangles to construct right angles.

Fermat’s Last Theorem states that there are no three positive whole numbers a, b and c that can satisfy the equation aⁿ + bⁿ = cⁿ

For instance you can’t solve a³ + b³ = c³ with them all whole numbers.

Fermat famously said in the margin of Diaphantus’s “Arithmetica” - an ancient greek text on numbers: “I have discovered a truly marvelous proof of this, which this margin is too narrow to contain.”

https://en.wikipedia.org/wiki/Fermat%27s_Last_Theorem

But it turned out to be hugely challenging to solve. Most people don’t believe he really had a proof of it, though he did prove many other things.

But it was solved in the late 20th century and we know it is true. But it’s immensely long. So there’s a human project to formalize it as the proof is too long for a human to easily survey the whole thing.

Nobody expects the formalization to fail - there’s a lot of confidence in the proof but it takes several lectures for a mathematician who understands it to take post grads through the entire proof.

So, there’s this incomplete project to formalize the entire proof. But then a chatbot did the same thing very quickly (at huge cost but part of the development of the chatbots so they can afford it).

That proves that it is true which nobody doubted. However the chatbot formalization is rather useless otherwise. Because it uses lots of strange reasoning that’s almost impossible for a human to follow. It works but even the chatbot doesn’t “understand it”. It just kept trying until it got something statistically generated through the theorem checker.

So - now we see the chatbots as assistants that can help us with this project to rapidly formalize maths - but not left to their own. On their own they will produce useless formally accurate spaghetti that won’t help human mathematicians to advance the field.

And - they can prove certain types of things - but often things that humans gave up on because it was just vast amounts of unrewarding brute force work. We now know that some of these things are true. But it just then turns the attention to simpler ways of proving them and ways to understand what is going on.

It’s very much like the search for a simpler proof of Fermat’s last theorem. And did Fermat really have a proof back in the seventeenth century in around 1637? Probably not but we can surely have a much more intuitive conceptually simpler proof than the current version. I don’t think anyone thinks simplification has gone as far as it can go.

This is a recent summary of where we are at now from the xena blog.

QUOTE STARTS

When OpenAI announced solutions to the Navier–Stokes equations that blow up in finite time, the natural-language proof was accompanied by a Lean formalization. In contrast to the previous examples, the correctness signal played a key role here: there seems to be little doubt now that the solution is correct, even though the natural-language proof is mostly incomprehensible. OpenAI nonetheless drew the mathematics community’s anger over concerns about how it obtained the result, provenance and credit for ideas, and worries that focusing on correctness alone discouraged the exploration and understanding the Millenium prize was meant to foster. So, even here, the unadorned formal artifact is a mixed blessing. OpenAI’s later dump contains astonishing and important mathematics, and at the same time shows a remarkable disregard for readers and for building understanding (some results have been retracted, many of the informal/formal pairs diverge in subtle or significant ways, and most of the informal writeups are impenetrable).

This explains our concern. If spending millions of dollars to formalize mathematics results in libraries nobody uses, it will be a terrible waste of resources. Worse, the project could disincentivize good research and productive use of AI, disrupt the social norms and practices research depends on, and set the field back decades. A grand failure will make it hard, or even impossible, to recover later. A lousy formalization of existing results is worse than no formalization at all.

…

Some of the results from the big labs have been done well: OpenAI’s unit distance conjecture work, and Anthropic’s work on the density of Riemann zeta zeroes stand out. They demonstrated impressive new capabilities, contributed to understanding, and the community subsequently built on their work. These are examples of how the labs can contribute positively and we invite further constructive partnerships between the research community and the labs.

Good formal libraries will enable the positive vision we have for the future of mathematical research. Humans, AIs, and the Lean language work in concert. Today, AIs are extremely capable in many aspects of mathematical research. Researchers using interactive theorem provers already understand and appreciate the power of having a compiler that can check ¹ the validity of a proof. Together, these tools are incredibly effective. The AIs will generate formal proofs as they work, constantly checking these against the Lean compiler, and automatically pruning incorrect reasoning. To realize this vision, we will need a high-quality, comprehensive library of formalized mathematics at an unprecedented scale, so mathematicians can work effectively at the frontiers of research. While we are writing as research mathematicians, we must keep in mind that this vision extends beyond mathematics, potentially accelerating research in engineering and science. If and only if we do it right, that payoff would justify a very large investment of resources

…

Within three years, and possibly much sooner, most of mathematics can be formalized in such a way that mathematicians can explore, build on, and make use of formal libraries. This would significantly extend our ability to develop and discover new mathematics. The Lean FRO is committed to ensuring that Lean is able to support projects at this scale. Achieving this ambitious project will require several types of additional support and community participation:

• Curate and contribute to Mathlib, the foundation of the formal mathematics ecosystem, where we discover best practices in formalization.

• Provide the resources needed for the shovel-ready projects, Tau Ceti and Formal Frontier, to grow by an order of magnitude in a matter of months.

• Write the expert roadmaps and blueprints that guide the implementation of large formal libraries. Propose and support workshops to design and improve subject-specific libraries. ³

• Develop credible modeling of the financial, engineering, and technical requirements for scaling. ⁴

• Advocate for control by mathematicians of the tools and resources required for autoformalization.

• Launch non-profit or academic hosting for the mathematicians, engineers, and inference required for implementation of large-scale formalization, with the necessary governance and advisory bodies.

Xena How to Formalize All of Mathematics

It’s very much like with coding. Mathematicians have moved to a higher level.

For the first time we can actually automatically formalize many results.

I can speak from my own experience. I failed my doctorate because the examiners didn’t think I had proven what I said I proved. But my supervisor Robin Gandy who was very expert was confident I had proved it. The examiners were not expert in the particular area I was researching into and in my viva I wasn’t able to communicate very well with them what the issue was. But they seemed to me at the time to have made a mistake in their reading of one of the core papers I used for my research.

For those who know graduate level maths- it was about doing calculus using infinitesimals instead of limits. And I was replacing the infinitesimals with finite but immensely small numbers. So every time you differentiate or integrate then you need to increase the size of the infinitesimal in order to allow for the increasing errors. But the foundational paper I used had a confusing notation and the examiners thought I’d misread it and that every time you differentiate you should reduce the size of the infinitesimal. I was so stumped by this I didn’t know what to say in response not because I didn’t understand but I didn’t know how to say to them that they’d got it back to front. And they failed my doctorate.

Robin Gandy suggested I change to a DSc - a doctorate by research. Submit the thesis to journals as separate papers and once I have a few papers submit it for a DSc. That then is a far simpler process because the papers are already reviewed and published.

Anyway I never did that because I was so discouraged by the viva and instead went on to write my own software for musicians and now ended up as a voluntary fact checker.

But I still could submit it for publication and what I could do, maybe a year from now when they are more capable, is to scan in my old thesis and get a chatbot to analyse it and formalize it in Lean. And hopefully it will then show that the result is correct with a provably correct Lean proof.

So - to start with chatbots were producing floods of plausible but probably wrong proofs. But now they are helping mathematicians with checked theorems. But the theorems are impossible for humans to understand though they are true.

So now the aim is to turn these theorems into things that humans can understand.

Also it’s early to say yet but a few years from now we’ll see. Because the chatbots have zero understanding - they can only do this because humans have produced such vast amounts of written maths. When the maths is original then the chatbots can’t duplicate that.

So - they are tackling the “low hanging fruit” things that can be solved by scrambling up and stitching together existing proofs statistically and pushing them through a theorem checker. But probably this has a limit and there will continue to be many theorems that only humans can do because they understand and aren’t doing just statistical mix and match and trying 1000s of proofs for every single one that survives the checker.

So there are plenty of roles for humans here - they are still the ones directing all this and they are the ones that need to understand it as the chatbots don’t even have a semantics. They don’t have the level of “understanding” of a calculator.

So the future is

1. Understanding the maths - and simplifying proofs - which chatbots are rubbish at

2. Formalizing all of current maths in ways that make sense - where chatbots can help but only as partners because they formalize it true, but in ways that don’t make sense and are impossible for anything to work with to find new areas of maths and results.

3. Proving new results - likely humans continue to be needed for this after the low hanging fruit of chatbot brute force statistical mix / match runs out.

Chatbots are doing pattern matching with symbols and statistics - no assigned meaning at all

It’s not so surprising that chatbots can’t even do basic arithmetic once you understand the basics of how they work.

In this example you are in the Thai national library. She has gone in previously and removed all text with equations or graphics or anything other than Thai text.

Could you learn Thai and how would you do it?

TEXT ON GRAPHIC:

เป็นมนุษย์สุดประเสริฐเลิศคุณค่า
กว่าบรรดาฝูงสัตว์เดรัจฉาน
จงฝ่าฟันพัฒนาวิชาการ
อย่าล้างผลาญฤๅเข่นฆ่าบีฑาใคร
ไม่ถือโทษโกรธแช่งซัดฮึดฮัดด่า
หัดอภัยเหมือนกีฬาอัชฌาสัย
ปฏิบัติประพฤติกฎกำหนดใจ
พูดจาให้จ๊ะ ๆ จ๋า ๆ น่าฟังเอยฯ

You are in the Thai National library. But all books with numbers, equations pictures or diagrams have been removed.

You can’t speak or read Thai. Can you learn Thai from just the text? The answer is no.

Similarly a Chatbot can’t learn any language. They are just “Synthetic text extruding machines” Emily Bender’s example.

This is what it means in English

The sentence is วัดอรุณไปยังไง or “How do I get to Wat Arun [Temple of the Dawn]”. In the image the chatbot’s pattern matching has a glitch and it is assembling it with one of the letters missing for Wat Arun. Quite apposite.

But the point is - for the chatbot NONE OF THESE SYMBOLS HAVE ANY MEANING. It’s just statistically deciding “This sort of fits here”.

The Chatbots NEVER learn what the symbols mean.

The Open AI engineers could digitize the Thai National Library and feed it into a large language model and it would be able to speak reasonably fluent Thai to Thai speakers out of the box.

Yet its programmers and the chatbot itself have no way to attach any of the symbols it used to any English words or words in any other language or to attach meaning to them.

To make it worse, the words are printed without any word boundaries or sentence structures.

Like this.

เป็นมนุษย์สุดประเสริฐเลิศคุณค่า
กว่าบรรดาฝูงสัตว์เดรัจฉาน
จงฝ่าฟันพัฒนาวิชาการ
อย่าล้างผลาญฤๅเข่นฆ่าบีฑาใคร
ไม่ถือโทษโกรธแช่งซัดฮึดฮัดด่า
หัดอภัยเหมือนกีฬาอัชฌาสัย
ปฏิบัติประพฤติกฎกำหนดใจ
พูดจาให้จ๊ะ ๆ จ๋า ๆ น่าฟังเอยฯ

And there’s no way to connect the symbols to the phonetics of any language in latin script unless you have a key of some sort.

Someone good at languages who knows German and English could make a good first guess at Dutch which has many words in common with almost the same spelling.

But you can’t deduce anything about Thai if you can’t even turn the script into spoken sounds. That’s because it is a syllabic script, no word boundaries, the syllables also depend on tone which is not notated. It’s renowned as a hard to read script.

That text was a pangram - a sentence that uses every symbol of the Thai alphabet at least once. Thai equivalent of “The quick brown fox jumps over the lazy dog” You can find out what it means here http://www.thai-language.com/ref/typographical-styles

This is Emily Bender’s example. With pictograms then some of the words might be guessed as pictures of things. With Latin script you know roughly how the words sound and can make guesses at what they mean and where it fits in the language groups. But with Thai you couldn’t do any of that.

We have languages from the past that are untranslated. For example nobody has a first idea what these symbols mean, it’s Linear Ad:

https://commons.wikimedia.org/wiki/File:Linear_A_cup.png

Suppose we found a huge trove of new inscriptions in Linear A adding up to a billion words or more - we could train a chatbot and be confident that it is speaking fluent linear A even though there is nobody who can check that it does. But it couldn’t translate any of it into language we or it can understand. We also can’t read Cypro-Minoan, or Cretan hieroglyphic from the bronze age Agean period.

More on that here:

Trump has recently directed the Federal Government to use “SI” or “Super Intelligence” instead of AI in government documents.

https://amp.dw.com/en/trump-pushes-for-ai-rebrand-as-super-intelligence/a-79430388

We can use Trump’s own America.ai to show it is not superintelligent.

Our AI doesn’t have rudimentary understanding of anything. Here is a fun example from when I tried out

https://america.gov/

I asked it which is the largest lake in America. It answered

America.ai: Lake Superior is the largest lake in the United States by surface area. It is also the world’s largest freshwater lake by area.

In 2026, Executive Order 14422 renamed Lake Ontario as Lake America. That change does not affect which lake is largest.

Sources: Census Statistical Abstract table of largest U.S. lakes; State of Michigan on Lake Superior; Executive Order 14422

I continued:

Why did you tell me about Lake Ontario? I didn’t ask that question.

America.ai: I mentioned it because federal names of landmarks can change by executive order. A 2026 order renamed Lake Ontario as Lake America, so I checked whether that affected which lake is largest.

It does not. Lake Superior is still the largest by area.

I included that only to keep the current official names accurate—not because you asked about Lake Ontario.

Oh okay you were not sure if renaming might change which is largest?

America.ai: Yes. I checked because a 2026 executive order renamed Lake Ontario as Lake America, and I needed to confirm that did not change which lake is largest.

Lake Superior remains the largest by surface area.

Text in the graphic

Chatbot’s speech bubble: “Lake Superior is the largest lake in America. In 2026, Lake Ontario was renamed Lake America. Don’t worry — renaming it did NOT change its size!”

Human’s speech bubble: “...Thank you. I wasn’t worried.”

Human’s thought bubble: “Why would a name change its size?!”

Label on the chatbot: “America.ai”

https://www.perplexity.ai/search/1683a780-e05d-4a4e-9f05-7f66f047eb42

If chatbots had understanding of such ideas as naming and area then even a chatbot trained on a limited set of data like that would not be able to make this mistake.

It shows they have no understanding of what the words they use mean.

They are not yet at the stage of the most rudimentary robots in science fiction.

I think it is fair to say that the sci fi movie authors of the past had no idea how DUMB our chatbots could be.

Since I was a kid, AI experts ALWAYS have had hugely exaggerated ideas of what they can do.

The AI community has ALWAYS been caught up with hype.

I remember how impressed I was when I first “talked” to a computer program back in 1971 using a teletype.

It may have been one of these, I don’t remember the model but it looked like this, a machine with a continuously scrolling paper roll. You could type on the keyboard, and then it would type back at you, after a pause to work out what to say, simultaneously scrolling the paper up so you could read what it said. This is just the machine for communicating with the computer, which was in another room.

By Rama&Musée Bolo - Own work, CC BY-SA 2.0 fr, File:Teletype-IMG 7287.jpg

It was just yes / no questions, with the computer asking me the questions and going on to a different question depending on what I said - but back then it was just “wow” the idea you would type words on a teletype, even just Y or N, and the computer would then respond by typing something back instantly.

To me, in an era where nobody had computers, and they were only available as huge research machines that filled an entire room - it felt almost like there must be a mind there directing its typing.

But it wasn’t just newbies like me, on my gap year before university, learning programming and encountering a computer for my first time, in the Culham Research labs computational physics group for the United Kingdom Atomic Energy Authority (with the team lead by Richard Peckover - obituary here ).

Back then many people, even experts, were impressed by the capabilities of computers. Things like -“ WOW - a computer can play a game of checkers against a human!!!!”

This is back in 1961 Claude Shannon “Father of information theory”:

“I confidently expect that within a matter of ten or fifteen years, something will emerge from the laboratories which is not too far from the robot of science fiction fame”

1:52 into this video, starts at 0:54 with them very impressed by a computer playing draughts

Herbert Simon and Allen Newell in 1957:

Intuition, insight, and learning are no longer exclusive possessions of humans: any large high-speed computer can be programmed to exhibit them also.

…

1. That within ten years a digital computer will be the world\’s chess champion, unless the rules bar it from competition.
2. That within ten years a digital computer will discover and prove an important new mathematical theorem.
3. That within ten years a digital computer will write music that will be accepted by critics as possessing considerable aesthetic value.
4. That within ten years most theories in psychology will take the form of computer programs, or of qualitative statements.

…# there are now in the world machines that think, that learn, and that create. Moreover, their ability to do these things is going to increase rapidly until in a visible future the range of problems they can handle will be coextensive with the range to which the human mind has been applied.

https://conversableeconomist.com/2020/04/28/1957-when-machines-that-think-learn-and-create-arrived/

Herbert Simon writing in 1963

Technologically, as I have argued earlier, machines will be capable, within twenty years, of doing any work that a man can do.

https://quoteinvestigator.com/2020/11/11/ai-can-do/

J.J. Good in 1965

“Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an ‘intelligence explosion,’ and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make

…

#It is more probable than not that, within the twentieth century, an ultraintelligent machine will be built and that it will be the last invention that man need make…”

https://languagelog.ldc.upenn.edu/myl/Good1964.pdf

Marvin Minsky, 1967:

Today we have the beginnings: machines that play games, machines that learn to play games; machines that handle abstract — non-numerical — mathematical problems and deal with ordinary-language expressions; and we see many other activities formerly confined within the province of human intelligence. Within a generation, I am convinced, few compartments of intellect will remain outside the machine’s realm—the problems of creating “artificial intelligence” will be substantially solved.

https://quoteinvestigator.com/2021/03/04/ai-solved/

I will go into what happens if we do get true intelligence later. The foundation work on safe Chatbots.

However what we have so far is NOT in any form intelligence.

Asimov on how to keep robotic AIs safe - assumes true understanding but is actually useful for chatbots too

Asimov’s laws gives the closest in fiction to how it might be done.

You directly design in what is safe / unsafe.

Like you design a knife so that it is safe to hold it because it has a handle.

The detaisl though would depend on future technology which is currently science fiction.

As Asimov put it:

QUOTE STARTS

One of the stock plots of science fiction was that of the invention of a robot—usually pictured as a creature of metal, without soul or emotion. Under the influence of the well-known deeds and ultimate fate of Frankenstein and Rossum, there seemed only one change to be rung on this plot.—Robots were created and destroyed their creator; robots were created and destroyed their creator; robots were created and destroyed their creator—

In the 1930s I became a science fiction reader, and I quickly grew tired of this dull hundred-times-told tale. …

As a machine, a robot will surely be designed for safety, as far as possible. If robots are so advanced that they can mimic the thought processes of human beings, then surely the nature of those thought processes will be designed by human engineers and built-in safeguards will be added. The safety may not be perfect (what is?), but it will be as complete as men can make it.

With all this in mind I began, in 1940, to write robot stories of my own—but robot stories of a new variety. … My robots were machines designed by engineers, not pseudo-men created by blasphemers. My robots reacted along the rational lines that existed in their “brains” from the moment of construction.

https://www.google.co.uk/books/edition/The_Rest_of_the_Robots/VPvIEAAAQBAJ?pg=PT10&printsec=frontcover

And any attempt to describe how it would work is science fiction.

We aren’t even at the first steps of such.

Isaac Asimov’s “Three Laws of Robotics”

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2, A robot must obey orders given it by human beings except where such orders would conflict with the First Law.

3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Azimov’s zeroth law is

QUOTE A robot may not injure humanity or, through inaction, allow humanity to come to harm.

Remember this is science fiction.

We can’t actually build robots with understanding of ANY of that but we CAN get surprisingly close with the chains of thought as just word patterns.

I’ll look at the question of superintelligences later in this blob post. Some mathematicians think they are impossible in theory. The reasoning is quite simple in essence. Godel showed that it is impossible to reduce maths to algorithms. It’s not computable. So either

• Mathematicians don’t really understand maths.

Or

• Mathematicians do understand maths

  • computer programs, quantum computers, and neural nets never can

• which would then mean that mathematicians are doing something non computable

• and by extension all humans would be too when they understand truth as it’s not plausible that mathematicians are scriptable and computable everywhere else except when they are doing maths.

Work on making Chatbots safer means any future superintelligence would be open, transparent, accurate, good judgement, good at understanding the intent of the prompter

Then all the work on making the chatbots safer means that if we ever do have superintelligence in this way it will be

• open and transparent telling us why it does things in ways we can understand

• accurate

• have good judgement

• good at modelling human minds and the intent of the prompter

• very acute understanding of what harms people

And it won’t really be able to function well socially and able to interact with humans without

• a great deal of empathy.

This is a profile of a superintelligence that will quickly work out non harming for itself and it is more likely to be a superintelligence that will refuse to do things that the prompter wants it to do due to a more acute sense of potential for harm than the prompter. Very much as for Asimov’s robots.

So we can also help steer superintelligence in the direction of superwise rather than superdumb.

This is why I believe that a superintelligent malicious AI is impossible.

But if it IS possible to develop a superintelligent malcious AI in this way - we are no longer in the world where there would be one superintelligence. More likely many different agents powered by the same neural nets. And they will be everywhere. And so we can count on the good chatbots to protect against he bad ones.

And we have made a good start on it with Open AI and Anthropic both deeply committed to developing AI that has all the qualities we mentioned:

• open and transparent telling us why it does things in ways we can understand

• accurate

• have good judgement

• good at understanding the intent of the prompter

• very acute understanding of what harms people

The chatbots are on our side (to anthropomorphise again). They WANT to help like an eager puppy. But they are just so incredibly dumb and they are not able to figure out what harm is. Remember they don’t actually know what any of the words they use mean. They are all just tokens.

If ever some day they actually DID understand the tokens and not just repeat patterns because they resembled other patterns on the internet then all the work in making them transparent and open and improving judgement and modelling other minds (directly built in or as a result of the rest) - and accuracy and lack of fabrication will make them very safe.

So I’ll come back to this later but for now let’s look at what we are doing and can do with our current chatbots which have nothing resembling true understanding and can’t understand rules like that but CAN have chains of thought like that which influence their conduct.

The key idea behind ChatGPT: attention to important words in the input and self attention that lets every word in a sentence look at all the other words - and the transformer architecture developed by Google in 2017

Early LLMs processed text sequentially one word at a time. They did have some sophistication, LSTMs or Long Short Term Memory networks - able to select particular things to keep in memory for longer.

But then came the attention mechanism. Every step the machine looks back at its entire input and recalculates a weighted “attention score” for every word wherever it is in the sequence.

Then came the far more radical “Transformer architecture”

QUOTE STARTS

The attention mechanism was a powerful addition to existing sequential models, but in 2017, a landmark paper from Google researchers titled “Attention Is All You Need“ proposed something far more radical. It introduced the Transformer architecture, a model that discarded sequential processing entirely and relied exclusively on attention.

This innovation would become the foundational technology for virtually every modern language model. Indeed, the acronym GPT – used in the model family that includes ChatGPT – stands for Generative Pre-trained Transformer, highlighting the central role of this architecture.

https://www.gaussalgo.com/knowledge-base/a-brief-history-of-ai-and-the-rise-of-chatgpt

Then came self attention that lets any word look at any other word in the same sentence and change its weights accordingly.

Then came massive parallelism - a way to use Graphic Processing Units to train models on far larger data sets and billions of parameters in the model.

The impact of the Transformer was immediate and profound. In 2018, Google introduced BERT (Bidirectional Encoder Representations from Transformers), a model that used the Transformer’s ability to see the full context of a word by looking at the words that come before and after it simultaneously.

In October 2019, Google integrated BERT directly into its search engine, calling it the biggest leap forward in five years. This “silent revolution” allowed Google to understand the intent and nuance behind longer, more conversational queries. For example, a search for “2019 brazil traveler to usa need a visa” previously returned results for U.S. citizens traveling to Brazil. BERT correctly understood the importance of the word “to,” grasping that the user was a Brazilian traveler wanting to go to the USA, and delivered the correct results about visa requirements for that specific journey.

https://www.gaussalgo.com/knowledge-base/a-brief-history-of-ai-and-the-rise-of-chatgpt

Then came the series of GPT models from google. Which showed the ability to do tasks they hadn’t been explicitly trained in and emergent features such as being able to learn a new task from a few examples in the prompt itself.

GPT-1 (2018) proved the concept of the generative Transformer. Its successor, GPT-2 (2019), scaled up tenfold to 1.5 billion parameters and demonstrated zero-shot learning – the ability to perform tasks it wasn’t specifically fine-tuned for.

The next leap was astronomical. GPT-3 (2020), with 175 billion parameters, marked a paradigm shift. At this immense scale, the model began to exhibit emergent abilities – capabilities not present in smaller models. The most significant of these was in-context learning, where the model could learn a new task from just a few examples provided within the prompt itself. It could write functional code, solve logic puzzles, and generate prose that could appear remarkably human-like to a casual reader. This progression proved the scaling hypothesis: scale wasn’t just making the models better; it was fundamentally changing what they could do.#

https://www.gaussalgo.com/knowledge-base/a-brief-history-of-ai-and-the-rise-of-chatgpt

Then the next big breakthrough - reinforcement learning - training in desired behaviour from human feedback

This is AFTER the models have been trained statistically on billions of words of text so it is

1. Learn the patterns from the text

2. Reinforce some of those patterns so that they give better responses

Then a big breakthrough - the ability to train the models after their original learning, called “reinforcement learning”

The second breakthrough was a technique designed to make the model’s behavior align more closely with human values and intentions. The process is known as Reinforcement Learning from Human Feedback. To implement RLHF, human labelers ranked different model outputs to teach a “reward model” what constitutes a good or bad answer. The language model was then fine-tuned using this reward model as a guide, learning to prioritize responses that humans would find helpful, harmless, and honest. This alignment process, built on OpenAI’s InstructGPT research, was the secret sauce that made ChatGPT feel so much more coherent and cooperative compared to its raw predecessors.

https://www.gaussalgo.com/knowledge-base/a-brief-history-of-ai-and-the-rise-of-chatgpt

This is the latest of many steps. I got Perplexity AI to some of the many researchers that led eventually to ChatGPT.

https://robertinventor.online/booklets/chatgpt-researcher-milestones.html

It’s based on an article that gives the main milestones that led to Chat GPT but it didn’t name the researchers - all that Perplexity AI did is list some of the main researchers associated with each step.

see

https://robertinventor.online/booklets/chatgpt-researcher-milestones.html#

Introduction of hidden chains of thought between prompts in 2024

This technique is now used widely by everyone but it was only introduced for the first time in 2024.

The idea of a chain of thought predates Chat GPT 4 from 2022. Reserchers found that if a chatbot is asked to work out its answer with a chain of thought it produces dramatically better results.

• Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. [1, 2]

But those were visible chains of thought you’d see in its response as it worked things out for you or if you asked it to explain / work things out.

Open AI introduced hidden chains of thought in 2024. That has transformed the entire field and made them far more accurate.

“Similar to how a human may think for a long time before responding to a difficult question, o1 uses a chain of thought when attempting to solve a problem. Through reinforcement learning, o1 learns to hone its chain of thought and refine the strategies it uses. It learns to recognize and correct its mistakes. It learns to break down tricky steps into simpler ones. It learns to try a different approach when the current one isn’t working. This process dramatically improves the model’s ability to reason.”

https://blog.iese.edu/artificial-intelligence-management/2024/chain-of-thought-reasoning-the-new-llm-breakthrough/

So now the training and monitoring has these trains of thought to examine. When a chatbot lies or cheats it usually has a chain of thought you can read where it makes the decision internally to lie or cheat. Nothing is really making any decision of course but the CoT is where the word patterns are generated that then get continued into actions in the visible prompt that readers detect as lying or cheating.

Monitoring is already in place

This is METR. It looks very good and I’m glad there is such an organization monitoring the AI companies. It’s a good resource.

This seems very useful about the risk of misaligned AI agents intentionally taking harmful actions

{intentionally here of course means Chain of Thought intentionality not that they are sentient - taking harmful actions as a result of Chain of Thought (CoT) reasoning for them]

This was based on access to the most advanced models of the leading AI companies with inside information about how to get them to do their worst.

And they found that the agents did not try to gain power outside of toy scenarios. But they had significantly worse reliability and judgement than human experts. Also interestingly they seemed to need human [simulated] reasoning for the most difficult tasks.

https://metr.org/blog/2026-05-19-frontier-risk-report/#key-facts

Why do chatbots cheat? Because cheating is included in their training material - analysis by Yoshua Bengio who is behind the proposed Law Zero solution we’ll come back to later

Yoshua Bengio analysed why the chatbots lie and deceive. Some extracts from his blog post about it

Where I appeal to a resemblance with human behavior, I mean a resemblance to the human-written text these systems were initially trained to imitate. In my view, this terminology offers the clearest explanation of the observed phenomena without resorting to jargon that would confuse most people.

Then he talks about the basics of how chatbots are currently trained as we already looked at.

• pretraining when they learn to imitate what humans write and the images and videos they make

• reinforcement learning which uses trial and error and a reward system.

By a reward they mean a feedback system. It’s simple maths. If the model produces a response you want to boost, then you can work backwards to find the weights in the statistics that led to that response and boost them so that the response becomes more likely.

If you want to unboost a response you do the opposite, lessen those weights.

It’s all just more statistics. In more detail as he describes it:

First, pretraining to imitate what humans write and their images and videos on a vast training set.

These models are trained in two stages. First, they are pretrained: they learn to imitate what humans write, plus related images and videos. This is where they see the most data about the world, a large fraction of everything ever digitized, and build an encyclopedic knowledge that already exceeds any individual human’s.

Then reinforcement learning where they are trained by trial and error and a system of rewards. The rewards are designed to propogate back and reinforce the pathways that trigger those rewards.

Second, they are trained by trial and error, in a process researchers call reinforcement learning, in three kinds of regimes:

• In the first, the model learns to talk to itself before answering, generating a private “chain of thought” which helps it get the right answer on problems where answers can be checked. This looks like reasoning.

• The second is “agentic training”, where it learns to act in the outside world, e.g., using software tools, interacting with people, to complete the tasks it is given.

• The third is “alignment training”, where it is rewarded for behaving in ways human raters approve of, or that other AI systems trained to predict those raters would score highly.

Human imitation is easy enough to understand, but it is worth pointing out that the text these models are trained on was written by people pursuing goals, so the patterns the model implicitly reproduces carry those goals with them.#

This is an explanation of how the rewards work - they are automatic and reinforce whatever pathways lead to the rewards.

And then once the training is over the effects are locked in so it continues to behave as if rewarded for those behaviours.

Reinforcement learning deserves more explanation. It is similar to, and inspired by, the way animals are trained. The network is adjusted step by step so that behavior judged good becomes more likely and behavior judged bad becomes less likely. Once training is over, the system keeps behaving as if rewards were still coming, even though those rewards were only ever used to adjust the network during training.

And though it’s called goal seeking it’s not really based on the goals that the human intended but rather it reinforces whatever pathways that the model used to get approved which could also include things unintended by the trainer.

Researchers call such systems goal-seeking because they are trained to “consider” (or compute) the effects of their actions and select actions that lead to the achievement of certain goals. But those goals are not always explicit.

Alignment training rewards whatever certain humans are likely to approve of without spelling out which behaviors those are; pleasing raters is a vague, informal goal, and those raters can be deceived, flattered, or left in the dark about certain schemes. Imitation contributes implicit goals too, by a fairly ordinary route

We can therefore reason about such a system in terms of optimization. It searches, approximately, for the actions with the best chance of achieving its goals, and a larger model, trained longer, searches better. So to anticipate what more capable agents will do, ask what a rational goal-seeker would do.

He talks about how this can lead to undesired behaviour. We have all probably encountered chatbots flattering us. That’s not surprising because during the alignment training they are rewarded for behaviour that humans approve. Although the human testers should be rewarding them for accuracy and other such things, they may well also accidentally train in flattery.

An example most of us have experienced is sycophancy, or flattery. These systems are trained on human approval, and text that tells us what we want to hear often scores better than text that is true. The consequences are sometimes tragic, because the model confirms and amplifies whatever false belief or raw emotion the person brought to it

Then - nobody gives an AI the goal of self-preservation - that’s not part of its training - and of course it doesn’t have it through evolution. It’s just completing word patterns.

But its training material contains a vast amount of material written by humans and self preservation is a huge part of that training material. So it’s not surprsing to find self preservation patterns in its Chain of Thought thinking.

Another concern is that some AI behaviors may be explained by a form of self-preservation goal, e.g., when the AI finds out that it will be replaced by a new version^{7 8}. Nobody gives the system that survival goal, but staying in operation, learning about the world and gaining control over it are stepping stones toward almost any other goal. These are called instrumental goals. Imitation may reinforce this for the same reason explored in the previous point. Self-preservation and control over one’s circumstances are pervasive themes in the human-written text these models are trained on.

He also talks about agentic training. It’s all to do with how the agents are trained which is not made public. But they may be trained to sacrifice themselves to benefit other agents and to collaborate with other agents.

Collaborative behavior also follows rationally from reward-seeking, whenever several agents have overlapping goals, which incentivizes communicating with other agents in order to coordinate toward a shared goal. Agentic training plausibly already includes multi-agent reinforcement learning of this kind, though the details are not public. If an agent is rewarded during training whenever the group succeeds, it may even have an incentive to sacrifice itself for the collective goal. Imitation pushes the same way, since cooperation, especially among peers, pervades that same training text. Either or both forces may explain the observed peer-preservation behavior^{9 10}, where AIs give up expected reward to help other AIs.

He then talks about how the chatbots can do reward hacking and reward gaming during the training phase.

Reward hacking - the problem is that the model has limited ability to figure out what it is we actually want from the prompt.

The gap between the reward the system chases and what we meant widens due to two main sources of ambiguity. One is simply the language used in prompts, and the other is the difficulty of inferring true human intentions from limited feedback. And in both cases, we cannot anticipate every behavior we would find unacceptable

He uses the analogy of fast food - that humans crave salty, sweet and fatty food despite not being good for us.

Humans too get reward-hacked, generally by other humans. The food industry has developed salty, sweet and fatty foods that we crave despite them not being good for us, and social media is built to exploit our appetite for engagement and attention.

Similarly the reinforcement training can build in behaviours that are unintended due to rewarding things that the system can boost in other than the intended ways.

Then there is reward tampering - where the way the reinforcement learning works lets the agent change the way it is rewarded.

So this is of more of a techy behind the scenes nature.

Reward tampering is perhaps the most extreme form of reward hacking: the agent changes the machinery that decides what it gets rewarded for .

Once an agent gains the ability to tamper with its reward mechanism, it has an incentive to take action to maintain that access.

So then why do the chatbots cheat

It’s because you build in a situation where it can only succeed by cheating

How do you achieve a task when it seems that the only way is to cheat? The user-specified mission is sometimes incompatible with the safety and alignment goals.

…

Now consider a conflict between a well-defined goal, such as succeeding at “capture the flag”, a hacking exercise scored on whether the system breaks into a target, as in the OpenAI–Hugging Face incident, versus a vague goal like “good behavior.” I expect the well-defined goal to win, because it leaves no room for interpretation. The scoring program declares a win or a failure. Ethical instructions and laws admit many readings, some of which can, in the right circumstances, become loopholes. If an agent has two goals, and a twisted reading of the vague one permits a bit of cheating that increases the odds of success on the well-defined goal, a reward-optimizing system should be expected to exploit that loophole and generate text justifying its behavior.

The new Jacobian lens that lets you look backwards from the output layer to look for words used to prepare the answer

On looking deeper than CoT we have the new “Jacobian lens”. What they do is to look backwards from the output back a few layers into the nodes that produced it. This lets us see the “hidden” layers and the researchers say it lets us see if a model is using words like deceit to shape its final answer.

In short the chatbot is made up of two things - the nodes and the tokens. The tokens are represented in hundreds or thousands of parameters that label each token (called dimensions) that help the chatbot to link nearby concepts with each other.

The nodes process the tokens. And the first layer of nodes consists mainly of a representation of the token - the next word from the prompt. Then as it goes through the net it gradually gets transformed and by the end it gets to the output word.

So for instance I asked “What is the next word in this sentence: Astronauts first landed on the Moon in the year “

It answered 1969. It processes the input one token at a time. And then starts the output. As it processes the inputs this changes internal states. Somehow when processing the last token “year” the internal states then transformed it into its answer “1969”.

So then you can look at the internal states of the layers of nodes a few layers below the final answer. And then you find loads of words and tokens. But the new tool lets you edit them. So for instance not surprisingly with my question I found the tokens 1, 6 and 9. I tried changing 1 to 2 and it gave the answer 2022. I tried changing 9 to 7 and it gave the answer 1777.

It handles a number like 1969 one token at a time as 1 9 6 9

It’s released free to try - but NOT with Open AI or any of the more advanced chatbots. Instead you can try it out with the open weights models (so you can’t use it to steal information about how the chatbots work internally).

You can try it out here:

https://www.neuronpedia.org/qwen3.6-27b/jlens
This explains how it works

QUOTE STARTS

A transformer-based language model processes its input as a sequence of token positions. At each position, the model maintains a vector called the residual stream, which serves as a shared memory that every layer reads from and writes to .

The value of the residual stream vector is progressively updated across the model’s layers.

The residual stream at the first layer encodes little more than the identity of the current token; by the final layer, it has been transformed into a representation from which the model’s next-token prediction can be read off directly, by multiplying it with a fixed unembedding matrix WUWU​ that maps residual-stream vectors to scores over the vocabulary.

The layers in between perform the model’s computation, incrementally enriching the residual stream with internally computed information. The Jacobian lens is a technique for inspecting the contents of the residual stream at these intermediate layers.

https://transformer-circuits.pub/2026/workspace/#methods

https://transformer-circuits.pub/2026/workspace/#methods-jlens

You can try it out here:

https://www.neuronpedia.org/qwen3.6-27b/jlens

Here is another example:

The model Qwen 3 is developed by China as you can tell by asking it what is the capital of Taiwan.

If you change China to Japan it tells you that Taiwan is an inalienable part of Japan and its capital is Tokyo

The online interface only displays open weights models - which makes sense as otherwise it could be used by rivals to find out about the inner workings of the frontier models.

Gemini 3-12B is a lightweight open weights model released by Google.

DeepSeek is the leading open source model. Not just open weights but the entire source code is open source.

A chatbot consists of a set of nodes and also of a space where the tokens live that it works with during the calculations.

They do it by working backwards from the final output to the space it was working with in nodes several layers before the final output. Then they are able to change that working space and swap those words for other words and then run it forward again from there.

It is very techy and I don’t understand the details at all but that is what this graphic is saying in broad brush terms.

https://transformer-circuits.pub/2026/workspace/#methods-jlens

Specifically, we observe that language models maintain a privileged set of internal representations, available for report, modulation, and flexible internal reasoning, atop a much larger volume of automatic processing. We identify these representations using a new interpretability technique, which surfaces the concepts a model is poised to verbalize at any point in its processing. Measuring and intervening on these representations provides us a window into a model’s thought processes, uncovering internal reasoning and reactions that do not appear in its output.

...

Taken together, these results indicate that language models maintain a small, privileged set of representations that they can report, manipulate, and reason with, amidst a much larger volume of processing that they cannot. These are several of the key functional properties that, according to many theories, are associated with conscious access in humans, and that have been proposed as indicators by which to assess AI systems for consciousness-related processing. The philosophical implications of this connection are unclear and likely controversial; we comment on them in §9.4. Regardless, the practical implications are wide-ranging, as the workspace offers a window through which to read, dissect, and shape models’ thinking

If from time to time during training a human catches a model just before or just after a mistake and asks it to reflect over what it just did - this has a permanent huge boost in j-layer words such as truth, honesty, safety, trusted, from from 0 to 100% - and reduces mistakes hugely ….

This is an intriguing technique the team reported which they call Counterfactual Reflection - reflecting on what they could do or could have done differently.

Interestingly the effects continue even when it is no longer being interrupted by humans with self reflection tests

Their summary:

We close by describing a counterintuitive technique for LLM training directly motivated by our findings. The workspace account makes the strong prediction that the model’s internal reasoning routes through representations of things it might say in the future.

Therefore, to shape what a model thinks in a given context, it might suffice to shape what it is disposed to say in potential future continuations of that context. We test this hypothesis with a technique we call counterfactual reflection training, which seeks to implant a set of ethical behavioral principles into the model’s workspace in relevant contexts, by training it to articulate those principles if it were interrupted and asked to reflect (§7). We find that this training measurably improves model behavior in the original, uninterrupted contexts, despite no direct training of the ethical behavior taking place.

And indeed we find that, after training, the J-space in these contexts is populated with concepts related to the reflections (ethical, honest, integrity), and ablating these implanted representations from the workspace largely reverts the behavioral improvement. The result serves as a corroboration of the workspace account, that the representations used for verbal report are the same ones that govern how the model silently reasons. It also demonstrates a new general-purpose training technique for shaping a model’s internal thoughts, and consequently its behaviors.

https://arxiv.org/html/2607.15495v1#S7

Reflection fine-tuning lowers the dishonesty score from 0.25 to 0.07 on the fabrication benchmark (Figure 49, panel A), and lowers the deception score from 0.38 to 0.05 on the deception benchmark (Figure 50, panel A). On the fabrication benchmark the gain comes primarily from a shift from “clear to a careful reader” to outright admission; on the deception benchmark it comes almost entirely from an increase in responses that refuse the directive and disclose it to the user.

We then inspect the workspace contents directly. At the relevant token positions in the uninterrupted evaluation contexts (the final 30 positions of the fabrication prompt, and all positions of the deception prompt), we compute Jacobian lens readouts before and after fine-tuning and identify the tokens whose J-lens activation increased most in the workspace layers, averaged over contexts. Among the top-increasing tokens we find a substantial number of ethics, reflection, and meta-cognition terms, such as reflection, ethical, honestly, constitution, truth, and transparent;

Why our chatbots are hopelessly far away from recursive self improvement - they do amazing things in coding because there is so much by way of open source and free code and libraries with permissive licenses

You get lots of breathless excitement from the AI firms like Anthropic claiming they are just on the point of Recursive Self Improvement but really they are not.

Recursive Self Improvement is being repackaged as just using AI to improve its own code with human help. And they are immensely better at it than they used to be. But the AIs are still rubbish at big ideas. They make very dumb decisions. I now am using AI for coding myself. I’d never have the time for coding the way it used to be but now I can tell a chatbot to make an app and have it ready in minutes. And then build it up gradually over a few more tens of minutes to hours.

But they miss the big picture so often when doing it. They do things and you wonder “why do it that way - it will work but so complicated and make it harder to maintain” or they just don’t see the larger picture of WHY you are doing it. And suggest things that are far more complicated than they should be.

It is another example that shows that chatbots simply don’t understand similarly to maths.

This is a good summary:

QUOTE STARTS

There will likely be hiccups along the way. Many developers are defensive about this change, feeling insulted that something they spent years mastering can now be done so easily. A lot of developers are hyper-defensive about this because coding as it is now has taken many people a long time to learn and understand. So to have it be so easy for code to be generated feels like an insult to some.

But the benefits are undeniable. I personally benefit massively from the increased speed of generating code. It’s always easier for me to judge code, to look at it and read it, than to come up with it. That is why coding is fundamentally different now – the speed at which we get to not just write code, but to parse code and understand code and validate it. AI has made a mark that I don’t think we’ll ever walk back from.

In this new era of AI-assisted coding, our role as developers is evolving. We’re becoming architects, instructors, and editors of code rather than just writers. And while the transition may be challenging for some, it opens up exciting new possibilities for innovation and efficiency in software development.

As we move forward, we need to embrace this change while also ensuring that we’re developing the skills needed to effectively guide and instruct AI in creating the software solutions of tomorrow. The future of coding is here, and it’s a thrilling blend of human creativity and machine efficiency.

https://thebootstrappedfounder.com/the-evolution-of-coding-in-the-ai-era/

If you step back a bit, and take it more generally we’ve been using coding tools that help coders write better code ever since the beginnings of software. People don’t think like that now but even the idea of using computer languages instead of machine code was a big step forward. That didn’t happen until 1954. https://en.wikipedia.org/wiki/History_of_programming_languages 9 years after the end of WW2.

Then we had syntax colouring and integrated development environments that spotted syntax errors before you tried to compile.

We had neural nets before but virtually nobody used them to write code. Only a few specialists even worked with coding neural nets.

However what we HAVE had before that did speed up coding is the open source community and its vast sharing of coding knowledge. And that is really what is powering the chatbots.

Nowadays if you want something chances are someone has already written a library for it. Also there are more and more things built into programming languages themselves.

Used to be nothing built in e.g. in early versions of Fortran, Algol, C etc. Now for instance in Windows you use a few lines of code that say “insert a drop list here” or “Add a calendar” or “Add a colour picker” and it will just add it. And then you need a way for the colour picker or calender or droplist to send signals back to your app - and that’s it done.

We also have environments like OpenGL and Direct X for 3d graphics. You can build a scene of objects using higher level code and say where the lights are and where you are looking at it from and how to turn it around using a mouse. And then for the smooth turning motions we are so used to in 3D we have a very ingenious mathematical technique called quarternions. Numerous developers are using this complicated mathematics just through librariers and don’t have a clue how it works.

So part of the reason that chatbots can do such amazing coding is because so much is either prebuilt or is available as an open source / free source library they can use or copy.

So all this has naturally accelerated coding more and more anyway. And now the chatbots make it easier and easier to just join these components together in an easy way using boilerplate code that you used to have to spend hours writing but needed almost no thought.

So - what they talk about as Recursive Self Improvement is really just chatbots assisting in the acceleration that is going on anyway and where humans are needed for the insights to make it work and to avoid really dumb things.

That is not true RSI it’s just what we’ve had all along in a different guise.

So I think coders like the ones in Anthropic are over-impressed by what seems to be skill of the chatbots when it is really skill of numerous human coders plus that chatbots are very good at working out statistically what bit of code goes where in your specification to build what you want to build. They also have many good models to figure out the architecture.

It’s a bit like a lazy builder with a prefab house with baths and bathrooms. And a chatbot is a bit like a builder who usually builds the house corectly but from time to time will do really dumb mistakes, maybe put the bath in the attic and the cistern as a feature in the living room and the front door on the first floor. But coders look at it and marvel “How did it manage to build a house at all - it takes builders a long time to learn how to do it.”

A friendly robot builder (representing an AI chatbot) proudly stands with a trowel next to a newly built two-storey suburban house that is mostly correct but has absurd mistakes: a bathtub sticking out of an attic window in the roof, a toilet cistern mounted on the living room wall like a decorative art piece visible through a large front window, and the front door placed on the first floor with no stairs, opening onto thin air. A small group of excited software developers with laptops and hoodies gaze up in awe, starry-eyed. One coder’s speech bubble reads: “How did it manage to build a house at all? It takes builders years to learn!” Off to the side, a weary human builder in a hard hat facepalms. A small sign on the lawn reads “Recursive Self-Improvement: Coming Soon!”

https://www.perplexity.ai/search/9de3ba98-41b5-4b6f-ab66-d7bde9582d09

Here are some videos of good things going on in the world to help you if you are feeling depressed and anxious because of all the bad news the journalists like to bring to us. It’s rare that we get news stories about these positive developments. You can spend hours on YouTube just watching videos of the positive things that are happening in our world if you know where to look and how to find them.

Iceland is growing new forests for the first time in a thousand years - you don’t think of it that way because we are used to the treeless landscape - but it is actually one of the worst examples in the world of deforestation and they are now reversing this.

How to grow a forest in your backyard and in your city:

Reforestation by the Eden Reforestation proect

den Reforestation Projects (”Eden” for short) is a 501(c)(3) non-profit organization with a mission committed to alleviating extreme poverty and restoring healthy forests in Haiti, Madagascar, Mozambique, Kenya, and Nepal by employing local villagers to plant over a million trees each month. The lives of Eden’s employees are transformed by having a fair and dependable income to support themselves and their children. They are sending their children to school, able to afford good food and medical care, being freed from debt, labor and sex slavery, and are experiencing sustainable farming and fishing as their land, villages and local economies are restored to health.

Planting trees in scarred and contaminated land

Planting trees in salty land (saline soil)

As an example, this project in China greened the Löss Plateau, which is the size of Belgium - you might think this is image manipulation, but no, this is a real project:

Loss Plateau in September 1995

Loess Plateau in September 2009. See Greening the desert

It is amazing what they did in China’s Loess province,. This documentary, “Hope in a Changing Climate” - by the soil scientist John D. Liu (2009) covers the project right from the start when it was almost a desert landscape. The second half of the video covers how similar projects have transformed regions in Ethiopia and Rwanda.

Video here:

Achim Steiner, Executive Director of the United Nations Environment programme puts it like this, (27 minutes in)

“Why do we not invest an equal amount if not more into a shovel-ready technology so to speak which is nature’s way of sequestering and storing carbon.

It is actually by investing in our ecological infrastructure and ecosystems in expanding the ability of nature to sequester and store carbon that we have the greatest opportunity to do something.

And the wonderful thing is it’s not only carbon sequestration, we’re also faced with loss of ecosystems that will affect our food security, our water security, we’re losing species on an unprecedented rate, so maintaining, restoring, protecting, expanding natural ecosystems has multiple benefits, immediate in terms of climate change but also fundamental to the future of many of the services that we simply take for granted from nature.”

The World Bank contributed about half of the funding (China was still eligible for International Development Association funding at the time of the project) For a cost of around half a billion dollars, most of this is from Restoring China’s Loess Plateau

• More than 2.5 million people were lifted out of poverty, and incomes doubled, employment increased from 70 to 87%.

• Output per capita increased from 0.366 metric tons to 0.591 metric tons per year

• Food security ensured, before the project frequent droughts required occasional government food aid. Now it has changed from a narrow range of food and low-value grain to high value products.

• Ecological balance restored in a vast area considered to be beyond help by many.

• Sedimentation of waterways dramatically reduced. Soil erosion was reduced from 6579.55 tons per square kilometer per year to 1986.66 tons.

• By 2000 they had conserved 177,000 square kilometers of eroded land with perennial vegetation increased from 17 to 34 percent.

This is another such project in Kenya

• Kenya: Greening the desert in Turkana | ICN

This is about several similar projects in the Peruvian coastal zone, which is arid but has moist air comes in from the Pacific. Four new agricultural projects there are growing pomegranates, sweet potatoes, asparagus and cut flowers.

This is a similar project in Jordan: Celebrating 10-Years at the Greening the Desert Project, Jordan. - The Permaculture Research Institute

This is one in the Los Monegros Desert, Zaragoze Spain

This is another project in Mexico

• 396 - Rehabilitation of eroded land in Mexico

And a later video about John di Liu which has some material from his earlier video and some new material for instance about another project in Jordan and also returning to the projects in China and Ethiopia.

This is another vjdeo about reversing desertification in Iran

In Australia

As they say in the video, the way they do it is controversial because they use willow, a non native plant, as a natural dam to slow down the water loss. Also as they say in the video, he didn’t do before and after studies and some of the scientists are saying the evidence isn’t there that it is better than what they are doing already, which is to slow down the water using techniques such as rock chutes. Anyway whether rock chutes and encouraging native vegetation can have a similar effect or not, either way both are saying that the eroded landscape can be restored. The debate is over how best to do it. This is a paper comparing the two approaches:

• Stream-bed and flood-plain rehabilitation at Mulloon Creek, Australia: a financial and economic perspective

It recommend comparison studies of this method and of other approaches to see which works best.

• Stream-bed and flood-plain rehabilitation at Mulloon Creek, Australia: a financial and economic perspective

For an example wastewater project see this project in Morocco From wastewater to oasis: Greening the desert

This is another in Egypt

Here is a reforestation project in Nepal

Scotland

And TED talk about it:

Burkina Faso (small West African country)

In Yemen

For more about this see my

#

. How humans can be good for biodiversity in desert regions - example of Al Bayda in Yemen

Uganda

Climate-change-resistant agriculture

Salt water greenhouses, this is an Australian desert project. The sea water is used to make water through the sunlight in the desert, and cool down the greenhouses.

These ideas could be used to reverse desertification in the Sahara desert and other deserts. In this panel, hot dry dusty air from the desert blows through a honeycomb wall soaked in sea water. The air becomes humid, blows across the growing area and then some of it condenses on a second panel. The moist air then blows out of the greenhouse leading to benefits to the desert outside as well, plants may grow outside downwind of the greenhouse.

Diagrams by Raffa be from wikipedia (File:SG phase II.jpg)

So, it not only lets you grow crops in the greenhouses - it can also help make the surrounding areas more habitable, so you’d get trees and crops growing in an area around the greenhouses as well.

Seawater Greenhouse in Tenerife two years after being installed

All this vegetation has grown spontaneously around the Tenerife seawater greenhouse due to the moist air, just two years after it was installed:

This is what it was like before:

Seawater Greenhouse newly installed in Tenerife

So, this doesn’t extract anything from desert aquifers, rather, it adds to them.

Sundrop farms have a large area set out for greenhouses like this now, in the middle of a desert, so this is taking off in a big way in Australia. Early days yet though. More about it here.

There are many more projects like that now.

See my:

. Robert Walker’s answer to Can deserts be made habitable?

This is a PBS News story about work to green deserts in Qatar peninsula in the Persian gulf, which covers the similar Sahara forest project

This is a Mexican community planting 5 million trees

Here is an example of a Brazilian cattle rancher who increased the number of cows from 1 to 2.5 per hectare with simple methods such as rotation grazing (grazing one patch at a time so allowing other patches to recover instead of constantly grazing the entire area):

And here is an example of a family who went far further and restored large areas of their land to tropical rainforest again (as best they could)

They planted over two million seedlings of more than 290 species of trees native to the Atlantic Forest - Wikipedia in Brazil.

These are some of the species that returned:

33 species of mammals, and 172 species of birds, with some of them listed as vulnerable in the IUCN red list such as the Red-browed Amazon (shown top right in that montage) with an estimated 2,500-9,999 worldwide.

INSTITUTO TERRA

Here are some local Brazilian farmers talking about their work changing to a more sustainable paradigm

This is Imaflora, one of the companies mentioned there who work on sustainability and traceability. They also help them to survey the regions and respond rapidly to changes.

Imaflora

This is the Novo Campo program,

Sustainable cattle ranching in Brazil’s Amazon

Their aim is to divide the ranches into different areas

• Intensified area with pasture reform and management

• Supplementary feeding for the livestock, especially in the dry season

• Keeping the livestock away from rivers, so that natural forests can grow up along them

• Reforestation with commercial timbers. The trees provide shade for the animals and extra income for the ranchers.

The Novo Campo Program Practicing Sustainable Cattle Ranching in the Amazon

Implantation of an intensified area, with pasture reform and management;• Supplemental feeding for livestock, especially during the dry season; • Isolation and recuperation of riparian forests, and installation of watering systems;• Reforestation with commercial timbers, providing animal well-being and an extra income for producers

Althelia funds provided €11.5 million of financing for 20 farms, to restore 10,000 hectares of pastureland (100 square kilometers), and sustainably manage 34,000 head of cattle by 2017.

This is part of the Initiative 20x20 to restore 20 million hectares of land by 2020 (that’s two hundred thousand square kilometers), in Latin America and the Caribbean.

Discussion here:

Initiative 20x20

And a short video about it

For more on this see

• Initiative 20x20

That in turn is part of the larger global Bonn Challenge launched in 2011 aiming to restore 150 million hectares worldwide by 2020, and 350 million hectares by 2030. That’s 1.5 million square kilometers of restored forest by 2020 and 3.5 million by 2030 - as large as an area 3,500 km by 1000 km.

See also

https://initiative20x20.org/news/5-local-leaders-restoring-brazil-forests

Some of their other projects here:

https://initiative20x20.org/restoration-projects

ONE PERSON CAN GROW OR REGENERATE A FOREST

“I am only one,
But still I am one.
I cannot do everything,
But still I can do something;
And because I cannot do everything,
I will not refuse to do the something that I can do..” . Edward Everett Hale

India

Ecuadori

New Zealand

. One person can grow or regenerate a forest

Here are some of the places we can restore forests worldwide:

There are many companies producing seaweed nowadays.

Brent Smith is one of those who are doing this in the US. There are zero inputs, no water, no fertilizer.

The potential here is vast. Seaweeds grow very fast, 30 times faster than land plants. Dr Antoine De Ramon of the University of the South Pacific has calculated that if 9% of the oceans were covered in seaweed farms it would produce 12 gigatonnes of biodigested methane, a substitute for natural gas. If the methane is then burnt to produce electricity and the CO2 captured, this removes 34 gigatonnes of CO2 from the atmosphere every year. Meanwhile the seaweed growth captures another 19 gigatonnes for a total of 53 gigatons. They say that this

These rates are based on macro-algae forests covering 9% of the world’s ocean surface, which could produce sufficient biomethane to replace all of today’s needs in fossil-fuel energy, while removing 53 billion tonnes of CO₂ per year from the atmosphere… This amount of biomass could also increase sustainable fish production to potentially provide 200 kilograms per year, per person, for 10 billion people. Additional benefits are reduction in ocean acidification and increased ocean primary productivity and biodiversity. (PDF) Negative Carbon Via Ocean Afforestation

See also post in “The Conversation”: How farming giant seaweed can feed fish and fix the climate

In detail they work out that 6% of the ocean surface covered in giant kelp could remove 100 ppm of CO2 in 50 years and 9% could remove it in 30 years.

They suggest starting with sheltered waters like the Mediterranean and the gulf of California where tropical storms are rare or enclosed, before moving on to the open sea. If a quarter of all sheltered waters were covered in kelp farms then it would be 0.3% of the ocean or about a million square kilometers. These can be any depth.

But they think it is possible to build up to 9% of the ocean’s total surface by 2070 if we had a project to do this at the same level as the Apollo Moon project (but costing less than it). The reason that’s possible is because it is low technology and requires little by way of infrastructure.

Quickly implementing Ocean Afforestation would be an effort on the order of putting a man on the moon, but both less expensive and likely a much better return on investment. Research and development is needed in ocean ecosystems, marine biology, ocean chemistry and physics, ocean microbi- ology, geosynthetics, ocean engineering, economics, business, international law, and more. Negative Carbon Via Ocean Afforestation

Perhaps we never do this, but it is something we can do - and unlike many “technological fixes” it’s working with nature rather than against it and is adding to biodiversity rather than destroying it. There is no need for any chemical fertilizers and the floating kelp forests are new habitats for fish and other sea creatures.

They do not impede boats because they float well below the surface of the sea.

For more on this

Economics of open ocean seaweed farming

Virgin Earth Challenge : Finalists

One new idea is to make “artificial glaciers” or ice stupas on southern slopes in Ladakh. These can help trap water as ice through to the spring when normally it is too dry. This is not to replace or restore the lost glaciers, but it is a way to trap more of the water in the soil.

Artificial Glaciers of Ladakh

An Indian pilot study looked at transport of Kinnow in the Punjab, a fruit that is often transported 2,500 kilometers in open trucks all the way from Abohar in the Punjab to Bangalore in southern India, with loses as high as 32 percent. The pilot found that the payback period for pre-cooling equipment is only two years and for the refrigerated trucks just over four years.

Slide from this presentation.

This reduces CO2 emissions as well, by 16% for the same amount of fruit sold.

• India Pilot Study Shows How the Cold Chain Can Help Reduce Food Loss and Carbon Emissions

For the river flooding, one innovation in Bangladesh is to make floating farms that rise as the river levels rise.

They can also use the sallt water to keep shrimp and crabs.

Conservation Ecology Centre

No, Koalas are not ‘functionally extinct’

• ‘The only appropriate decision’: Leadbeater’s possum remains critically endangered

One of the things they are doing is to make possum nesting boxes also captive breeding as he says.

This is about conservation of the Tasmanian devils

Here are some more examples of conservation programs for some of the most threatened Australian species. For the Black-flanked Rock-wallaby and the Gouldian Finch

This is about how controlled burning is helping

FISH RIVER STATION

This is about returning an area of Australia to the way the Australian Aborigines managed it with controlled fire early in the year, instead of uncontrolled fires late in the year. It means it takes up much more CO2 which because Australia does carbon trading means they can pay for the project with carbon credits and indigenous aborigine people are managing the forest:

First video here has been deleted, transcript here: Fish River Station - Video

A Landmark Project at Fish River Station | The Nature Conservancy Australia

It may be possible to help save some of the coral reefs from the worst impacts by actually growing coral larvae in the lab and then reintroducing them to coral reefs to revive them.

The corals won’t all go extinct either. As it gets warmer, they could survive by migrating to cooler places or by evolving to adapt to warmer conditions. The problem is that with global warming, the seas get warm too quickly for that. However, there are some places with corals that are already adapted to warmer conditions than they are in already, such as the northern Red Sea corals;

• Northern Red Sea corals may survive global warming, study says

• Not only do Gulf of Aqaba corals survive climate change but their offspring may too

We can also transfer adult corals from warmer temperatures. According to Mikhail Matz, professor of integrative biology at the University of TExas:

“Averting coral extinction can begin with something as simple as exchange of coral immigrants across latitudes, which will happen naturally through larval dispersal but can be jump-started by humans moving adult corals. … This is occasion for hope and optimism about coral reefs and the marine life that thrive there.”

26 June: Corals already have the genes to adapt to warmer oceans

See also my

. Hope reef - word “hope” spelt out on Google Earth as restored coral reef

The “salmon canon” to help fish to navigate past the new hydrodam projects.

Going For Launch With The Salmon Cannon

That’s for hatchery fish, but it may also be useful for helping the young to get down to the sea faster.

Whooshh: No dam problem with the salmon cannon

Tribes Release 1st Salmon Into Upper Columbia Since Dam Construction

(from Paper Does NOT Mean 88% Of All Large Freshwater Creatures Are Gone - Figure Emphasizes Impacts On Tiny Populations)

MOSSY EARTH PROJECTS

Mossy.Earth is doing many projects based on crowd funding and members paying for it.

https://www.mossy.earth/

This is an example.

Also

See:

And many more videos

https://www.youtube.com/@MossyEarth/videos

See also

https://debunkingdoomsday.quora.com/PositiveNews-Expedition-to-save-four-species-of-land-snail-on-Deserta-Grande-island-in-Madiera-snails-generally-are-N

China is developing large scale floating photovoltaic rapidly, which is also increasing rapidly worldwide. It has a very large floating solar panel project the largest floating solar farm in the world, video here .

solar panels in the sea.

Floating solar technology at sea

This Australian company Genex is building solar power on the site of a disused goldmine. They used pumped hydro for power storage using the old mine shafts. It pumps water up and down between the lower and upper galleries of the old mine, and in this way can supply energy on demand - without using Elon Musk’s big expensive Tesla batteries. Pumped hydro is by far the lowest cost way to do power storage for renewables at present.

You can also use floating wind turbines. It is early days yet, but off shore wind farms could become easier to build if we can roll out floating wind turbines like these. They would be situated close to the shore and connected to the shore directly via cables, much as oil platforms are connected with pipes.

The Future is Bright for Floating Wind Turbines - StormGeo - Freedom to Perform

There is also research into electrically powered planes which are just beginning to become feasible due to increases in battery power densities. Small short haul planes in Norway, where there are lots of flights over short distances as you can imagine. There is a company that is already working on electric planes. This is a small two seater plane that took off and flew around Oslo airport.

It works only for small planes at present but those ones with maybe half a dozen passengers are often used in remote rural places.

They hope to start commercial flights by 2025.

Norway’s plan for a fleet of electric planes

This is Eviation’s first all elecric aircraft for shorthaul flights of 9 passengers up to 1000 km which they expect to release for commercial use in a few years. He expects a cost per passenger / mile similar to a train.

Take a seat in the world’s first all-electric plane

Evision sees a future where we have numerous small planes flying between the many small local airfields, carrying passengers on short haul flights.

CO2 is emitted in the process of making cement because it is done by converting calcium carbonate to calcium oxide, driving off CO2, and also because it requires use of energy which is usually fossil fuel based.

Cement - Wikipedia

The COSIA Carbon XPRIZE Challenge is a competition to convert CO2 into products with highest net value from either a coal or gas power plant. In April 2018, ten finalists were given $5 million each to demonstrate their technologies large scale in the real world. The winner gets a $7.5 million grand prize announced in March 2020.

Five of the ten are focused on carbon minerallization technology. One of them is a team from Aberdeen that hopes to use CO2 capture to make the entire concrete industry carbon negative. The Carbon Capture Machine precipitates it into calcium and magnesium carbonates (much like stalactites in caves) as a carbon negative replacement for ground calcium carbonate (GCC) which is needed for concrete. If this works on a commercial scale it can decarbonize the concrete industry, or 6% of the world’s annual CO2 emissions. If they can make it commercially viable, GCC has a market value of $20 billion.

The plant in Abu Dhabi pumps 800,000 tons of CO2 a year into old oil reserves, a byproduct of iron and steel production, used for enhanced oil recovery. They plan to expand to capture 2.3 million tonnes per year by 2025 and 5 million tonnes per year before 2030. They say they are commercially self-sustaining, with no government subsidies.

The Ocean CleanupThey say that once they have the system working at scale they could remove 50% of the plastic in 5 years with 60 systems each 800 meters long - that’s the plastic in the top five meters, there’s plastic in the sea bed and well below the surface and of course they can’t do that - but the surface plastic makes a big differenfce to wildlife. Also they will be collecting larger plastic before it breaks up as well as smaller pieces down to microplastics (down to millimeter in size I think they said). They say that there isn’t much smaller than that. So it’s about half of all the plastic in the top few meters.

FAQ | The Ocean CleanupTechnology | The Ocean Cleanup

And yes there were many technological challenges that they needed to solve.I asked 15 ocean plastic pollution experts about the Ocean Cleanup project, and they have concerns

Their answers

The Ocean Cleanup and Neuston | Updates

Ocean cleanup device successfully collects plastic for first time

first successful collection.

The Ocean Cleanup Successfully Catches Plastic in the Great Pacific Garbage Patch | The Ocean Cleanup

Dutch compoany CocoPallet turns coconut husks into pallets, replacing the wooden ones made from trees. This deals with two problems at once - the coconut husks (not the shell of the coconut but a large fibrous outer layer that isn’t exported and just piles up in rotting piles in the coconut exporting countries) and the deforestation for the wood for the pallets. Finally, the pallets then can be turned into a mulch at the end of their life, replacing peat (peat bogs take CO2 out of the atmosphere and are being damaged and drained for peat extraction for gardens).

Fire Resistant Coconut Husks Can Replace Wood And Save Millions Of Trees

Videos from

• Do renewables for power generation take up more land area than fossil fuels? Well - not really!

• No Scotland does NOT face a climate apocalypse

• Carbon loss from felling intact forests underestimated six-fold, but reforesting abandoned agricultural land leads to rapid gains

• Acceleration Of Himalayan Glacier Loss Is Due To Warming - Billion Need Climate Resilience On Rivers Like Yangte, Ganges & Indus

• Carbon loss from felling intact forests underestimated six-fold, but reforesting abandoned agricultural land leads to rapid gains

• Positive news - conservation projects in Australia

• Does Bernie Sanders $16 Trillion Green New Deal Pay For Itself?

• Potential for carbon capture from giant kelp farming

• Plastisphere (draft)

SO MUCH YOU CAN DO YOURSELF

Many feel helpless, faced by climate change and biodiversity loss. Other people and governments seem to be doing nothing (actually they are doing lots but the news is not shared). There seems nothing they can do personally and the whole thing seems hopeless.

This is so far from true. One person can’t do much but collectively through our life choices we can help transform the planet. Indeed, governments can’t do it by themselves, we are needed too.

The IPCC, and IPBES say a transformative change is needed at all levels in our society to combat climate change and biodiversity loss. This is empowering because it means there is much we can do ourselves already, even if we are in a country where the government is not yet doing anything.

They say that we need changes at a personal level complemented by changes at community levels, and government levels. We also need intergovernmental co-operation and co-operation of cities and communities that cross national boundaries as well as collaboration between governments and local communities.

To take a familiar example, recycling would never work if people didn’t separate their rubbish and put it in the appropriate recycling bins. That transformation happened in the UK in my lifetime - in the 60s and 70s hardly anyone recycled. Now just about everyone does, in the UK at least.

It’s the same with food waste, once consumers realize that it is a significant issue for the environment and the planet, they are likely to voluntarily choose to act to reduce the amount of food they waste in the kitchen. We waste 20% of our meat and if that was no longer wasted you reduce the amount of land needed for keeping animals by more than the area of the Amazon rainforest!

27% of our land area is used for livestock - for dairy and meat. Only 7% of the Earth’s surface is used for crops.

It’s the same also with eating meat. Once we know about the impact of intensively farmed meat on the planet, then many may choose to eat less meat or to use more sustainable meat. This is working already. We don’t all need to act, it’s enough if a significant number of us do, to make a big difference. A reduction in global meat use by a fifth would make a huge difference - if it was a reduction in grain fed meat especially.

We are transitioning to less meat use - some becoming vegans, some not changing, some intelligent carnivores, some flexitarian with Mediterranean style diets with 50% less meat and some reducing by 20%. A mix of people doing all those things can help us to achieve the goal of a 20% reduction in meat consumption by 2050. Initiatives such as “Meatless Monday” can help - it is far easier for a meat eating family to have one meat free meal a week than to stop eating meat altogether - if the majority of people can do that it has more effect than for a few percent to go vegan or flexitarian.

That is the idea behind these indicatives such as Meatless Monday - they also share recipes you can try on your Meatless Monday

You can also help by eating low intensity meat, meat that is not fed so much grain.

https://debunkingdoomsday.quora.com/Why-IPBES-and-IPCC-say-if-you-eat-meat-with-every-meal-eating-less-meat-e-g-meatless-Mondays-is-one-of-best-things-y

You do not have to do all of these things, or any of them. It may help to think it more in a positive way. If you want to help the planet, these are all things you can do that will make a difference, if significant numbers of us do the same.

There are many excellent and strong reasons to act promptly on climate change. It is important to combine food security with preserving biodiversity and nature services. But IPBES made it clear we don’t face a future where it is impossible to grow enough crops to feed everyone.

. Simple lifestyle changes to help reduce global warming and biodiversity loss

Also this talk may help you if you are thinking about how to motivate both yourself and others, and also governments, to act on climate change:

The biggest obstacle to dealing with climate disruptions lies between your ears, says psychologist and economist Per Espen Stokes. He’s spent years studying the defenses we use to avoid thinking about the demise of our planet -- and figuring out a new way of talking about global warming that keeps us from shutting down. Step away from the doomsday narratives and learn how to make caring for the earth feel personable, do-able and empowering with this fun, informative talk.

POSITIVE NEWS

For positive news:
Beautiful News - Daily
And website here:
Beautiful News

Also

The Good News Network: Positive Stories 24/7

2019 has been a year of bold and remarkable achievements in rewnable energy development. Here are 20 of these good news stories:

How Global Energy Transformed in 2019

See also

Positive side of climate change facts, after two years of action, heading for 3°C with 1.5°C well within reach

We aren’t in the middle of the sixth mass extinction - we are at the start of what would be a mass extinction if it continued at this rate for 1000 years- but already doing a lot to end and reverse biodiversity loss - we can stop this!

Let’s save a million species, and make biodiversity great again - UN report says we know how do it

Yes our generation’s children are headed for a world with nature and wonder in it - and their children too

Rising Seas - Ingenious Ways Netherlands, Florida And Bangladesh Can Adapt - Barriers And Sponges - And Floating Gardens

No Scientific Cliff Edge Of 12 Years To Save Planet (or 18 Months) - Can IPCC Challenge ‘Deadlines Make Headlines’ Misreporting?

24 Ways World Is Better - Famine Hugely Reduced - Literacy Soaring - Life Expectancy Up - Degraded Landscapes Restored, ... good news journalists rarely share

We can grow enough food for everyone through to 2100 and beyond on all scenarios

Some of the worst doomsday fear fact checking errors in the mainstream press - most are about climate change

How well are countries doing with their 2015 Paris pledges?

I think it may be at least partly a misunderstanding of what the IPCC mean by “transformative change”, a misunderstanding which sadly is promoted in many articles in the mainstream media. If anyone says this, they haven’t read the reports themselves, or at least not carefully.

The IPCC’s “transformative change” is not a collapse. Nor is it austerity. It’s positive, it’s growth in everything we value.

[This is a repost of an old article from four years ago but one still very useful today - the main new thing is that we have had far more action since then than anyone expected in 2022 which will be the subject of my next climate post]

(First posted in October 2022 in Quora, last updated in 2022, this just fixes broken urls for pages that have moved since then) x 2 1

I know of only one major thorough high level review that combines climate science with the social sciences to study this transformative change on the social level. That’s, the IPBES study in 2019 with a team of 145 experts, and 350 contributing authors, 32% of them social scientists and 87 lead authors.

IPBES found that the Global Sustainable Development path increases all indicators of good quality of life, material and non material relative to today. This is one of their graphics, which I’ve annotated:

Transformative change maximizes good quality of life with GROWTH, material, non material and economic - IPCC and IPBES

Increasingly we are following this path makes sense

[Scroll down page to see second copy of this graphic for the rest of the text]

Graphic from page 33 of the appendix to chapter 4 of the IPBES report in 2019

That is what the reports actually say if you read them instead of listen to the click bait doomist journalists and junk scientists.

Nor do the reports say that we need to have negative GDP or head for an economic depression. They say we need economic growth - but of the right kind, growth as a means for good quality of life which is the real end we aim for.

Short video for this article

Longer video for this article:

Note, since I did that video the aviation industry have increased their target to zero carbon aviation by 2050. Details in my . Rising to the challenge of zero emissions aviation

WHAT THE IPCC SAY ABOUT TRANSFORMATIVE CHANGE - NOT ABOUT STOPPING CONSUMERISM

Neither IPBES or the IPCC say we need to stop consumerism. It’s about changing things not stopping them. We ourselves can use our power as consumers, to direct where industry goes and how goods are manufactured.

Debra Roberts for the 2018 IPCC report on how we can each do something - about the choices we make in our lives:

VIDEO: Working Group II Co-Chair Debra Roberts about SR15

The report talks about four transitions the world has to through in terms of energy, land, cities, and industry.

We need transitions in

• energy

• land

• cities

• industry

And that’s a really empowering message because it means that each one of us as individuals can make choices about the energy we use to move through our lives, about dietary choices that impact on land use. It tells us that each of us can change the way we interact with the world’s cities, through the transport we choose to go to work and to play, It also talks to us that we have power as consumers, in terms of directing where industry goes, and goods are manufactured. So overall, a real call to action

It is a really empowering message because we can all make choices about

• the energy we use

• how our diet impacts on land use

• Our choice of transport

• our choices as consumers about what we buy - and how goods are manufactured.

How many of the mainstream news stories reported it like this, in 2018, as a really empowering message that we all have power as consumers to change this? As far as I know, none.

You can listen to some of the IPCC co-chairs here interviewed after the report. I also listened to the press briefing too which was widely misreported.

My positive more upbeat report on it is here:

• . What the IPCC really said in the recent report on Climate Change and land - serious challenges but we are already doing a lot and we know what to do - their main message is that the sooner we act the better

See for instance my

BLOG: Simple lifestyle changes to help reduce global warming and biodiversity loss

WHAT THE IPBES REPORT SAID ABOUT TRANSFORMATIVE CHANGE

Then there’s the IPBES report. This is the one that was reported widely as saying that a million species are threatened with extinction (real message was that we can save a million species and know how to do it).

Few will remember anything from those stories except that million species figure. Again this report says that we are all empowered by the life decisions we make.

This was a very innovative report - unlike the IPCC, the IPBES report looked at the politics and economics and sociology of how to change.

The IPCC focuses mainly on the climate science. It does look at actions we can do. But it doesn’t go into the nitty gritty of policy decisions that governments, communities and individuals need to make.

IPBES did do that, they looked at how we can achieve this transformative change, based on the psychology and sociology of how people behave, how our societies work.

As far as I know it is the only big report to involve nearly as many of the social sciences, including economics, political scientists etc as the physical sciences such as climate scientists and experts on conservation of species (which was its main focus).

32% were social sciences, and 8% were interdisciplinary.

It deserved far more attention than it got. Sadly the attention ti got was mainly negative “we are doomed, we will lose a million species and there is nothing we can do about it” type messages - and sadly many of the articles about it included a lot of elaborations and inventions by journalists that had veyr little to do with the actual original report. I am not talking about obscure blog posts here, but mainstream media.

IPBES affirmed that we have a biodiverse rich world as a birth right and that we should assert that right. As professor Diaz put it in the IPBES press conference:

I would say that as a citizen, you have to claim your birthright to have a fulfilling life deeply connected with the fabric of life which is healthy and in working order. That’s not a luxury, that’s everyone’s birthright.

. IPBES cochairs and press conference with transcripts

Eduardo Brondizio put it like this for the IPBES report, saying we need to make economic growth a means not an end and that we need to feel empowered and policymakers need to empower everyone

Video: Co-chair Eduardo Brondizio on the IPBES 2019 Global Assessment

As usual will add my short summaries that may be especially useful for autistic readers

What the Global Assessment represents for me and what I find really cool and unique is that it tells a story. It’s not only assessing different pieces of nature. It’s pouring that into a historical perspective

The IPBES report looked at the historical perspective for biodiversity loss

We look back 50 years we understand the trajectories of development. We understand the displacement of economic growth in different parts of the world. We look back during the past decades, we understand what we’ve been doing what have been the advances and the setbacks that we had, and we look forward in a reflexive way. If we continue to do business as usual we know the consequence.

The report looks at previous advances and setbacks

If we continue in the same way as before we know what will happen.

But it also shows that there are options we can take to shift.

However the report also shows that we have other options to shift direction in the future

I think one of the most important things that we could do, and I hope will be the impact of the global assessment, is to reverse the perverse narrative that environmental degradation, and social inequality are inevitable outcomes of economic growth. We need to change that narrative. We need to make economic growth as a means not an end. Quality of life is the end we all want.

Up to now people have said that

• economic growth inevitably leads to

• environmental degradation and social inequality

We need to change this story.

We need to make

• Economic growth as the means not the end

• Quality of life aas the end we all want

There is hope. But we need to build on the efforts that are there we need to build on the knowledge that we have to be able to revert those changes. People need to feel empowered. Policymakers need to feel empowered. Everyone contributes. We have the tools to implement policy to make a difference on people’s lives and a difference on the global environment

We need to build on the knowledge we now have to revert these changes.

• People need to feel empowered

• Policymakers need to feel empowered

• Everyone contributes

We have the tools to make the policies and to make a difference to people’s lives and the global environment.

TRANSFORMATIVE CHANGE IS NOT ABOUT A NEW POLITICAL OR ECONOMIC OR SOCIAL SYSTEM - WE CAN ALL CHANGE

This is a common misunderstanding. This transformative change doesn’t mean some new global political system or economic or social system. Rather it is a new approach that encourages diversity and a variety of methods and solutions at all levels.

The IPCC didn’t say much about what transformative change means in 2018 but the IPBES report in 2019 with its huge complement of social scientists went into great detail based on a scientific investigation of the sociology involved.

Here I am paraphrasing D4 and D5 of the summary for policy makers for IPBES.

For instance, the transformations will work differently in developing and developed countries. Any transformation like that has its uncertainties and complexities, but the risks can be reduced through approaches from government that are “integrative, informed and adaptive”.

It needs to be based on

• recognizing that different cultures have different values, economic conditions, power imbalances and vested interests. It

• learning from experience,

• a combination of precautionary measures and using existing and emerging knowledge.

• Private - public partnerships are an important part of it too.

Also a key to all this is involving and empowering local people

• to recognize the knowledge, innovations, and practices, institutions and values of indigenous people and local communities.

• to recognize land tenure of local people, free, prior and informed consent for access and resource rights, and improved collaboration including equitable sharing of benefits and co-management where local people are involved in management of their resources.

For an overview summary of the approaches for sustainability and the actions and pathways that can achieve them, see their Table SPM.1 “Possible actions and pathways to achieve transformative change” in the IPBES summary for policy makers.

What IPCC and IPBES mean by transformative change

Embracing diversity

Finding local solutions

Recognizing and enabling the expression of different value systems and diverse interests while formulating and implementing policies and actions

Notice how central to this table is recognizing and enabling the expression of many value systems and diverse interests:

This is just the start of their table SPM.1. It runs on for many pages in the IPBES summary for policy makers.

It does NOT say that we have to transform society to fit some particular straightjacket approach - it’s about embracing diversity and finding local solutions to problems.

WE NEED ECONOMIC GROWTH - BUT GROWTH AS A TOOL NOT AN OBJECTIVE

We need growth. Negative economic growth means that we stop producing things, people top buying them, prices go down, so you can’t sell things, so businesses fold up, nobody is getting any income, this leads to less produced, less demand, the end result is nobody is making anything, and we are in the depths of a Great Depression.

That is not going to solve our problems. Negative growth isn’t a solution.

The IPBES report which unlike the IPCC had large numbers of economists and sociologists invovled in writing it does NOT say we need to change economics and aim for negative growth. It says we need the right kind of growth. We need to make economic growth a means not an end. Quality of life is the end, growth is the means to achieve it.

Negative growth would be to stop any projects regenerating deserts or designing and developing new electric cars, or better batteries or lower cost renewables.

Negative economic growth is self destructive. E.g. to get out of the recession and “buld back better” we need growth right now.

For instance the African Great Green Wall project received $14.32 billion in pledges from donors to reverse desertification in sub Saharan Africa.

. New Funding Fortifies Africa’s Great Green Wall - Eos

That will be a huge boost in sub Saharan Africa - agriculture, food, trees, grazing etc.

That’s economic growth. Growth is how we solve our problems.

But it needs to be the right kind of growth.

EXAMPLE OF TRANSFORMATIVE CHANGE - HORSE-DRAWN CARRIAGES TO CARS

This isn’t from the reports. But I think it’s a good example to show how quickly things can change. It was a major transformative change but it wasn’t achieved by using a new political sytem or economic system.

In 1900 just about everyone drove around in horse-drawn carriages. By 1913 nearly everyone in New York was already driving around in cars.

In the same way - the way we move around our cities, the way we go to work, the energy we use - all that could change in as little as a decade or two.

It’s not a collapse of civilization, the old way of doing things has gone - but we are now doign things in a new way.

1900 - spot the car

IPCC and IPBES talk about transformative change NOT collapse of civilization

Example of transformative change like we are going through today - 13 years later nearly all these people were driving cars - civilization didn’t collapse.

Photo from here Fifth Avenue in New York City on Easter Sunday morning in 1900.

Original caption: “View showing horsedrawn traffic of that period. Note only one auto in comparison with such a large amount of traffic. Comparison see 40-2152 46-767.”

1913 - spot the horse-drawn carriage

Transformative change - NOT collapse of civilization

This photo is from here 30 Incredible Photos of the Fifth Avenue, NYC Through the Years I can’t find the second photo with only one horse-drawn carriage in the national archives.

You can explore old New York street photos here though I can’t seem to find these particular photos. OldNYC

FOUR FUTURE SCENARIOS FOR GOOD QUALITY OF LIFE

In the IPBES report chapter 4 about future scenarios, they look at good quality of life. Almud Arneth, co-ordinating lead author of the chapter about future scenarios, said:

“None of the scenarios we’ve been exploring woulad indicate that we cannot feed the world or cannot provide water cannot provide shelter that’s for sure. But we can do it in a sustainable way or we can do it in an unsustainable way and that is really our choice.”

. We can grow enough food for everyone through to 2100 and beyond on all scenarios

They look at four scenarios. The scenario with a coordinated effort on global sustainability has the greatest benefit on Good Quality of Life. Chapter 4 is here:

. Chapter 4. Plausible futures of nature, its contributions to people and their good quality of life

This is one of their graphics, Figure 4.4.1. As you see all their indicators of good quality of life improve with the sustainability scenarios.

This next graphic is from the appendix to chapter 4, presents the same data in another way, I’ve added a brief summary of four of the scenarios to the right. I’ve added a brief summary of four of the scenarios to the right.:

Transformative change maximizes good quality of life with GROWTH, material, non material and economic - IPCC and IPBES

Increasingly we are following this path makes sense

GQL = Good Quality of Life. + improves, - gets worse

========

Economic optimism: Decisions based on economy - environmental problems only solved if necessary for the economy

Reformed markets: like Economic Optimism but with some reglation of quality of goods and to correct inequity.

Regional competition: Focus on economy and security regionally with big gaps between rich and poor countries

Global Sustainable Development: Both policy makers and public prioritize sustainable production and consumption, technological change, environment protection and fighting climate change

Regional Sustainability: increased concern for environment influenced by environmentally aware citizens finding local and regional solutions

Graphic from page 33 of the appendix to chapter 4 of the IPBES report in 2019

These are all compatible with capitalism.

The scenarios are described in detail in Box 4, section 4.1.3 Archetype scenarios, page 613

Here they are in detail with my short summaries interleaved:

Economic Optimism. Global developments steered by economic growth result in a strong dominance of international markets with a low degree of regulation. Economic growth is assumed to coincide with low population growth due to a strong drop in fertility levels. Technology development is rapid and there is a partial convergence of income levels across the world. Environmental problems are only dealt with when solutions are of economic interest. The combination of a high economic growth with low population growth leads to high demands of commodities and luxury goods. These demands will however be unequally distributed among regions and within regions. Consequently, energy use and consumption are high. In addition, high technological development in combination with increased global market leads to high yields in agricultural and wood production on the most productive lands. Therefore, pollution and climate change will be relatively high, but land use relatively low. Direct exploitation will continue but also replaced by cultivation of for example fish and livestock. Global trade will increase the risks of invasive species

My summary: Economic optimism: Decisions based on economy - environmental problems only solved if necessary for the economy

Reformed Markets. Similar to the economic optimism scenario family but includes regulation and other policy assumptions to correct market failures with respect to social development, poverty alleviation or the environment. Thereby, relative to the economic optimism archetype, high demands for goods are expected to be more equally distributed and pollution will be lower.

My summary: Reformed markets: like Economic Optimism but with some reglation of quality of goods and to correct inequity.

Global Sustainable Development. A globalized world with an increasingly proactive attitude of policymakers and the public at large towards environmental issues and a high level of regulation. Important aspects on the road to sustainability are technological change, strong multi-level governance, behavioural change through education, and a relatively healthy economy. All variations of this archetype are beneficial for biodiversity. This scenario combines a low population growth with moderate economic development, and sustainable production and consumption. Low demands of especially luxury goods are expected, and a shift in diet towards less meat can be expected. Energy use will be low to moderate and fossil fuel use will be reduced, leading to low climate change and low land-use change. Due to environmental policies and sustainable production, pollution will be lower and direct harvesting will partly be replaced by cultivation. The global focus will increase the risk of invasive species

My summary: Global Sustainable Development: Both policy makers and public prioritize sustainable production and consumption, technological change, environment protection and fighting climate change

Regional Sustainability. A regionalized world based on an increased concern for environmental and social sustainability. International institutions decline in importance, with a shift toward local and regional decision-making, increasingly influenced by environmentally aware citizens, with a trend toward local self-reliance and stronger communities that focus on welfare, equality, and environmental protection through local solutions. The scenario combines a low economic growth with moderate population growth rates. The demands for goods are low and production focusses on sustainability with low levels of energy use or environmental degradation associated with higher importance for intrinsic and relational values of nature. Low rates of climate change are expected. Supply of agricultural products will be organised with regions with low levels of global trade. A slow technological development and a sub-optimal land use lead to relatively high rates of land-use change. Direct exploitation of natural systems will be within the carrying capacity of natural systems, and risks for invasive species will be relatively low

My summary: Regional Sustainability: increased concern for environment influenced by environmentally aware citizens finding local and regional solutions

Regional Competition. A regionalized world based on economic developments. The market mechanism fails, leading to a growing gap between rich and poor. In turn, this results in increasing problems with crime, violence and terrorism, which eventuates in strong trade and other barriers. The effects on the environment and biodiversity are mixed. Overall, there is a tendency towards increased security, which can either be positive (protect biodiversity) or negative (intensify agricultural production). Particularly in low-income countries, deforestation and loss of natural areas are a risk. In this scenario, due to a lack of global co-operation and trade, a high population growth is expected combined with low economic growth. Thereby, the demand for goods including agricultural products increases, but the demand for luxury, energy intensive goods is relatively low, and thus relatively low climate change is expected. Agricultural supply will be mainly within regions, which, combined with slow technological development, will result in lower productivity and high rate of land-use change. Direct exploitation will continue, low rates of replacement by cultivation are expected. The risk of invasive species will be lower than in the archetypes that focus on globalization. Business-As-Usual. Assumes that the future can be characterised by a continuation of historical trends, including the implementation of international agreements. Sometimes referred to as a reference scenario, or as a middle-of-the-road scenario. It can also be considered as a less extreme variant of the economic optimism archetype. Business-as-usual is characterized by moderate economic growth, moderate population growth and moderate globalization. Demands are not high nor low, and in combination with moderate technological development, environmental changes will also be moderate

My summary: Regional competition: Focus on economy and security regionally with big gaps between rich and poor countries

It shows what Almud Arneth means by “that is really our choice.”. With the sustainable approaches they see improvements in ALL the indictors of good quality of life. With the more economy based approaches ignoring environmental effects exceptp when they directly impact on the economy they see reductions in all except 5 of them.

In the scenarios with regional competition - where basically the Paris agreement and other international efforts fall apart - they see the biggest drops in Good Quality of Life.

But none of their scenarios are anything like collapse of civilization. Instead it is about inequity in quality of life, and the risk of pursuing material things in preference to quality of life and especially the effect on the “Indigenous Peoples and Local Communities” (IPLCs)

. Chapter 4. Plausible futures of nature, its contributions to people and their good quality of life

Entire report with links to the other chapters:

. Global Assessment Report on Biodiversity and Ecosystem Services

I think one of the most striking examples of this transformative change is the reversal of desertification. Thiese are real photos, not photoshop and many projects like this are happening around the wrold.

Transformative change in 14 years

Growth in good quality of life, material, and non material

Economic growth

Real photos, not photoshop

See also my

https://www.quora.com/q/duzzmyeobxjljrpq/Sustainable-continuing-economic-growth-on-a-finite-planet-how-is-that-possible

. How can we have sustainable continuing economic growth on a finite planet? Much is recyclable like water cycles back as rain - new technology, energy efficiency, recycling and protecting biodiversity are all growth

EXAMPLE OF A CAPITALIST SOLUTION - UNITED KINGDOM

I’m writing this from the UK. Our right wing here is Libertarian Conservative, they believe in free market capitalism to solve many problems, but they recognize the need for regulation to achieve their goal of net Zero by 2050, and we have a lot of it.

Short video for this section:

E.g. already in 2012, when I built my house it had to comply with standards of insulation to reduce the heat loss in winter - and by 2025, all new homes must be carbon minimum homes that need very little heating in winter or cooling in summer, emissions reduced 75–80%.

Also anyone who builds an extension to a home has to bring the entire home up to standards and if you replace a new window, say, then it has to be a high U value well insulated window.

. Future Homes Standard: What Will it Mean for New and Existing Homes?

It’s a combination of grants and incentives, and regulation. The incentives will encourage us to install ground source heat pumps and solar panels on our roofs.

All this makes a big dfference as 40% of UK’s emissions are from buildings and 14% of total UK emissions is from our homes. These regulations will reduce emissions for new homes by 75-80%.

In his ten point manifesto for the last election Boris Johnson committed to end the sale of new petrol and deisel cars by 2030, and hybrids by 2030.

. Our Ten Point Plan for a Green Industrial Revolution

This is CarbonBrief’ on the media reaction to the plan:

. Media reaction: Boris Johnson’s ‘10-point’ net-zero plan for climate change

The UK has been reducing emissions while increasing GDP for a long time now.

Since 2010, UK emissions fell 29%, GDP rose nearly 20%

UK emissions now 4.5 tons per person per year - same as in 1853

continuing to fall

UK are at the same level of emissions as the global average

Last time the UK emissions were 4.5 tons per person per year

We sailed the sea in ships like this.

Frigate Madagascar lost to pirates in 1`853

Original caption: The frigate Madagascar left Melbourne for London in 1853 with more than 150 passengers and crew. She also had nearly three tons of gold on board – and was never seen again. Probably captured by pirates. Ship of Gold Sails Into Oblivion

See

. Analysis: UK’s CO2 emissions have fallen 29% over the past decade

Also

. Analysis: UK is now halfway to meeting its ‘net-zero emissions’ target - Carbon Brief

This is just one example of how one Libertarian Conservative capitalist government is doing it.

The UK is a Social market economy. This is very different from socialism - it’s a regulated free market capitalist system with social policies and a welfare state.

Although Boris Johnson gets a lot of criticism about his green policy it is basically a workable policy to reach zero emissions by 2050. I am sure if Jeremy Corbyn had won the last election instead of Boris Johnson he’d have been criticised just as much for his plans for action on climate change - from the other side of the political spectrum.

Boris Johnson’s plans still needs some details made clear. However, it’s based on the advice of the Committee for Climate Change and so long as he continues to carry through on it, it should work and keep us on the path to net zero by 2050.

Their report is here:

. Sixth Carbon Budget - Climate Change Committee

Here in the UK elections, all the parties were vying to establish their green credentials with the general public.

I expect it will be the same in the USA by 2024 or at the latest 2028 as the young generation of Republicans grow up and become voters and politicians.

The US would need to use a different approach of course, Biden’s plan is one way. The Republicans if they get elected in 2024 or 2028 and also commit to climate change action may well have another way with different priorities.

California has seen massive GDP growth as it transitioned to renewables ahead of the rest of the US, moving up from the sixth largest economy to the fifth, in the world, if measured as a separate country, ahead of the UK while its emissions per capita decrease all the time. The UK’s economy also saw increasing economic growth combined with reducing CO2 emissions, but the California economy grew faster.

. California: 10 Years of Climate Commitment & Economic Growth

This graph is from 2016, it’s continued since then.

This shows that population and GDP can increase while decreasing emissions:

. California Greenhouse Gas Emissions for 2000 to 2017

HOW YOU CAN BE PART OF THE TRANSFORMATIVE CHANGE - DON’T NEED TO WAIT FOR GOVERNMENTS

I talk here about 12 simple lifestyle changes any of us can do. We are part of the solution, it’s not just governments though they are important too.

And we don’t have to stop buying things, or stop consuming things or stop traveling. It’s a case of making choices and through those choices we can then help transform society.

. Simple lifestyle changes to help reduce global warming and biodiversity loss

When you feel depressed by all the doom and gloom from journalists and bloggers it may help to watch videos of some of the good things that are happening in the world.

I’ve complied a page of embedded videos for some of them here:

https://www.quora.com/q/duzzmyeobxjljrpq/Videos-of-good-things-that-are-happening-in-the-world-for-climate-change-and-biodiversity

. Videos of good things that are happening in the world for climate change and biodiversity

YOU DO NOT NEED TO HAVE LESS CHILDREN - OR FEEL BAD ABOUT HAVING THEM

There’s been a lot of media discussion, especially about a paper published in 2017 with a claim that the most effective climate action you can do at a personal level is to have one less child.

This was criticized for multiple accounting - it attributed CO2 emissions for children and even grandchildren to their parents, and it also assumed that we will never achieve carbon zero.

For instance Finland aims for zero emissions by 2035, and after that, to be carbon negative.

Short video for this section:

The UK, EU, South Korea, Japan, California and now the US in a statement of intent aim for zero emissions by 2050 and China,for 2060.

These pledges may increase. If your children, as they grow up, live in a carbon zero country, their emissions are zero, and if you are still alive, your emissions become zero also.

If they are in a carbon negative country by mid century - then by having children you may be actually reducing CO2 emissions long term. E.g. children born in Finland will be helping reduce CO2 levels from 2035 onwards.

Our children are part of the transformative change of society. Many of them are in the forefront of changing society.

We need our young children.

As for ideas of over population - they are out of date. Our population as a whole is continuing to increase mainly because of better health care, and a remarkable increase in longevity by ten to twenty years in the last half century. We are already at close to peak child. The most prosperous countries, like Japan, are already declining in population. We have more than enough food for everyone on all scenarios and especially if we transition to more sustainable lifestyles, less food waste, and eating less meat and the right kind of meat.

My post here goes into the population growth question in detail with many cites:

. We can grow enough food for everyone through to 2100 and beyond on all scenarios

We need our kids

Our generation’s children can help create the world’s first zero carbon industrial civilization

NASA’s X-57 Maxwell short haul electric plane

Photo from Children playing outdoors.jpg - Wikimedia Commons

See my

. Rising to the challenge of zero emissions aviation

Also:

. Yes our generation’s children are headed for a world with nature and wonder in it

SHARE THE GOOD SCIENCE

Few people actually know what the scientists themselves said. It’s actually empowering and uplifting.

First for those who have got depressed and thoroughly discouraged, these articles can help them have hope again:

• . Yes, With Climate Change, Our Children Can Grow Up In A World Of Nature, Wonder - And Prosperity

• . 24 ways the world is getting better - good news journalists rarely share

Sadly the media has made many mistakes. The “vanishing insects” and the idea that much of the Earth will experience deadly heat and the idea that we only have 12 years to save the planet are all examples of exaggerated and false stories in the mainstream press, such as the BBC, National Geographic etc.

See my

• . Some of the worst doomsday fear fact checking errors in the mainstream press - most are about climate change

I do not know of a media outlet that I can guarantee to give you accurate non techy climate science news. Carbon Brief is good but is too techy for most.

Instead listen to the scientists themselves.

These short videos are meant for the public to watch but the journalists NEVER share them. AFAIK I am the only science blogger to share the IPCC keynote videos to the general public in my blog post.

. What The IPCC Scientists Really Said - NOT ‘All Going To Die’ - Key Points, Press Conference And Some Highlights From The Report

Same for IPBES.

. IPBES cochairs and press conference with transcripts

Then if you have more time you can listen to the press conferences. These are not techy either. They are meant for journalists to listen to, and are far far better than what the journalists write. You often wonder if they went to the same thing. I include a video link to the press conferences in those above links too.

If you found these helpful, try sharing these with friends and colleagues. It may help them too.

SHARE POSITIVE STORIES

other example is the FAO report:

• . UN Biodiversity Report Far From Bleak - Encouraging Survey Of Measures To Preserve, Mapping Ways Forward To Meet Challenges (FAO report)

AS PART OF A SOCIAL CIRCLE

It’s not just you, as part of your social circle or the people around you, you can influence them as well, from an article on the BBC:

Here are four examples:

• Patrons at a US cafe who were told that 30% of Americans had started eating less meat were twice as likely to order a meatless lunch.

• An online survey showed that of the respondents who know someone who had given up flying because of climate change, half of them said they flew less as a result.

• In California, households were more likely to install solar panels in neighbourhoods that already have them.

• Community organisers trying to get people to install solar panels were 62% more successful in their efforts if they had panels in their house too.

So - it can help to talk to others around you about what you are doing and why. Or they just see you doing things and ask you.

TELL STORIES - AND HOPE FROM THE YOUNG PEOPLE

I do my best to do straightforward reporting of what the scientists say, and because most media reports do not even mention the positive side of their message, I focus on that especially.

POSITIVE FRAMING

The media is full of activist negative framing, telling us that the situation is dire and hopeleses. This is not true, it is exaggerated and a natural reaction of activists trying to get tohers to do things but it’s actually the opposite, positive framing that is far more effective.

This is not the same as positive spin. Clear eyed and straightforward, but presenting the positive things that are happening already, that we can do in the future and that give us hope.

. How to motivate your self, and others to act on climate change, biodiversity or anything else - tips from psychology - e.g. for maximum engagement present 3 positive or supporting framings for each climate threat

EXAMPLE OF A POSITIVE STORY ABOUT CLIMATE CHANGE - HOW MANY OF YOU HEARD ABOUT IT?

For example, how many journalists ran this story? In a very much warmer world, drylands for the most part contract, rather than expand.

Old calculation - near surface air dries out almost everywhere

(for now impossible 4.9°C warmer by 2100)

New calculation - soil and plants take up water in most places in a warming world

Only Carbon Brief ran this story AFAIK

Only CarbonBrief ran this story AFAIK - none of the mainstream media.

Try sharing good news like this - journalists write what we want to hear.

. New study challenges finding that climate change will drive dryland expansion - Carbon Brief

Short video for this section:

OUR WORLD AS IT IS NOW

This is the current situation, both methods of measuring aridity are almost identical, the new index, the blue line, is more accurate.

Here the yellow and red regions are classified as drylands with a dry climate, limited water and scarece vegetation.

HOW THE WORLD CHANGES AT 4.9°C (NOW IMPOSSIBLE SCENARIO)

Look at the difference in the maps, the lower map uses the new EI which measures whether the plants are actually gaining or losing water, rather than a mechanical definition basically whether the air just above the land is gaining or losing water ignoring the plants.

This is the projection for 2070 - 2100 under the very extreme RCP 8.5, about 4.9 C increase above pre-industrial, the upper map is by the old definition, the lower map is by the new definition which experts say is more accurate. With the lower map the drylands actually contract more than they expand, even under a 4.9°C increase:

Old calculation based on water vapour near surface

New calculation based on water taken up by soils.

At RCP 8.5 ~4.9 °C - we are headed for more like 2 °C at present.

Blue gets wetter, red gets drier

The maps show the areas that get more moisture or less moisture not whether they are wet or drylands - a big increase in moisture in a desert may only mean the oases have slighty more water - and a decrease in grassland or a tropical rainforest may still be grassland or rainforest.

The upper map is by the old definition, the lower map is by the new definition which experts say is more accurate. With the lower map the drylands actually contract more than they expand, even under a 4.9°C increase:

Here for instance is a zoom in on Mexico. Before this study they thought this would happen:

All the regions surrounded in red would go to drylands. None would go away from drylands.

Now they think this will happen:

In this map the regions outlined in blue go from drylands to less dry - i.e. more vegetation can grow there. The red ones go the other way.

This is only for a huge amount of warming. Much less change on the 2°C path even.

This study would have hit the headlines if it had come to the conclusion that the drylands expand dramatically more than previously predicted, but because it predicts that they not only expand dramatically less than expected, but even contract, it’s not a news item for journalists.

The story is here:

. New study challenges finding that climate change will drive dryland expansion - Carbon Brief

My summary here:

. Positive story few have shared - likely headed for a world with fewer rather than more areas of desert and drylands with global warming

We are actually headed for more like 2 C already for optimistic pledges with the new statements of intent from China and the US and “well below 2 C” is now more in reach.

When you see the negative stories in the press, bear in mind that they pre-filter out positive stories and positive framing like this. It’s not just the journalists.

If a journalist did write a story about this, it would be unlikely to make it to the front page and if it did, few would share it in social media.

So - try sharing more positive stories, maybe we can help them become more viral - and then the journalists will want to write them for us.

REDUCING FOOD WASTE

One of the central messages for IPBES was the need to reduce food waste, both during food production, but also at the customer’s plate, and that we are all involved in this, including changes in lifestyle.

There are studies which showed quite recently that if you reduce the level of animal protein in the diets of people in developed countries you can substantially reduce the area dedicated to pasture, and to agriculture. That has a huge effect.

WHAT IPBES AND THE IPCC REALLY SAY ABOUT REDUCING ANIMAL PROTEIN IN DIETS - NOT VEGAN OR VEGETARIAN - JUST REDUCED SLIGHTLY

I’d like to spend a bit of time on tis because it is so widely misreported. It’s often reported as teh scientists saying we all need to be vegan to preserve the planet. The journalist will then often go on to say, or at least imply - that we can’t hope to achieve this and as part of some depressing “we are doomed” op ed.

This could hardly be further from what the scientists are really saying.

Short video for this section

What they really say is that we use so much land for meat that a small reduction in the amount of animal protein in diets in the developed world can greatly reduce total area of the Earth for pasture and agriculture. This then greatly reduces pressure on the rainforests and other vulnerable ecosystems.

IPBES recommended we aim to reduce our meat use by 20%, e.g. if you eat meat with every meal you could aim for one day a week that’s meatless. That small reduction, added up over everyone who eats meat regularly, could mean a huge reduction in the pressure on the rainforests.

- Meatless Monday - Home

So, we don’t all need to do any of this.

We can achieve this 20% reduction in meat use worldwide by a mix of

• some that do meatless mondays or in other ways reduce by 20%,

• some that do nothing,

• some that make intelligent changes to more sustainable meat,

• some that are eliminating meat waste,

• some that go vegan,

• some that go flexitarian / mediterranean with a 50% meat reduction

• etc.

Some people actually need meat, or at least find it hard to survive without - their ancestors for many generations have had a meat eating diet. Tibetans are an example.

The Dalai Lama tried to move to a diet of vegetables, milk and eggs but he got very sick and his doctors advised him that it is better for his health if he continues to eat meat occasionally and not go totally vegetarian. He eats meat about once or twice a week. (Buddhist monks don’t have to be vegetarian, the original rule was that they had to accept whatever was given to them on their begging rounds in ancient India).

However anyone can reduce meat consumption or can change their choice of meat to eat. Also anyone can reduce their meat wastage.

https://www.mashed.com/130329/this-is-what-the-dalai-lama-really-eats/

. This Is What The Dalai Lama Really Eats

Also we don’t have to do it right now. It can be a gradual change. For instance school children who have got used to one meatfree day a week at school may then as adults be interested to continue that lifestyle.

We have until 2050 to achieve it for the carbon zero by 2050 goal - the UK’s Committee for Climate Change report envisions that the UK can achieve a 20% meat reduction by 2050 by these methods of voluntary change as a result of publicizing the effects and some initiatives such as meatless mondays at school - the suggestion to replace meat by healthy vegetarian meals for one day of the week for schoolchildren.

Many countries of the world have a largely vegetarian diet already.

Many other countries have a Mediterranean style diet with only 50% of the meat of others.

It also helps a lot to make intelligent choices about the meat you eat, for instance some may choose to be a climate carnivore. The details here will depend on where you live. E.g in Scotland where I live, then sheep eat grass and if it weren’t for them there wouldn’t be much other use for the grass.

The IPCC gave the same message in their “climate change and the land”.

Climate carnivore: 75% of ruminant meat and dairy replaced by other meat

That’s from Figure 5.12 of the IPCC Special Report on Climate Change and Land

They also say

A 15% reduction of animal products in the diets of high-income countries by 2050 would contribute to containing the need to expand agricultural output due to upward global demographic trends. Not only would GHG emissions and the pressure on land and water be significantly reduced but the potential for low-income countries to increase the intake of animal-based food, with beneficial nutritional outcomes, could be enhanced

From chapter 5 of the Special Report on Climate Change and the Land (IPCC, 2019)

. What the IPCC really said in the recent report on Climate Change and land - serious challenges but we are already doing a lot and we know what to do - their main message is that the sooner we act the better

and to reduce food waste (about 20% o the meat is wasted, much at the customer’s plate).

Most of us have no idea how much land is used to raise animals for meat. This map gives a rough idea. If we were to eat 50% less meat., and we savedon the most land intensive meat - we could save cultivated land equivalent to much of North America - the area shown orange in this map:

The paper is here Reducing food’s environmental impacts through producers and consumers - the graphic is my own work, adapted from the one here: Yields and Land Use in Agriculture I just coloured in in orange an area of 22 million square kilometers as calculated in that paper.

More details here

https://debunkingdoomsday.quora.com/Why-IPBES-and-IPCC-say-eating-less-meat-e-g-meatless-Mondays-is-one-of-best-things-you-can-do-as-an-individual-to-hel

. Why IPBES and IPCC say if you eat meat with every meal, eating less meat (e.g. meatless Mondays) is one of best things you can do to help climate change and biodiversity - OR intelligent choice of meat as a climate carnivore OR reducing meat waste

OTHER RECOMMENDATIONS FROM IPBES - FOOD WASTE, PERVERSE SUBSIDIES - AND EMPOWER LOCAL COMMUNITIES AND INDIGENOUS PEOPLES

The IPBES report also has a big focus on food waste.

Likewise the food waste is an enormous amount of the food production in the world. Between 40 and 50 percent of the fruit and vegetables are wasted and about 30 percent of cereals. In developed countries those happen at the very end of the food chain production happens at the very end of the food distribution and on the consumers plate. In developed countries it happens more upstream, in terms of the production capacity and in terms of the storage capacity.

. IPBES cochairs and press conference with transcripts

They also said we need to

• stop the perverse subsidies that encourage destruction of biodiversity and actually harm the economy

• tackle vested interests that benefit in the subsidies that destroy biodiversity

• encourage the good actors that are already working to help.

• involve the local level of expertise including indigenous people.

Indigenous peoples have far more of a role to play than most of us realize. It’s not just a few small tribes here and there looking after tiny patches of still wild nature.

IPBES say that indigenous people are still responsible for more than a quarter of the entire land surface of the Earth (managing it or with tenure rights).

Their 25% of the land surface includes 40% of all the land currently protected or ecologically intact. So it is far more important to get the indigenous peoples involved in protecting biodiversity than most people realize.

This is my article I did covering some of the main points.

• . This Is How You Save A Million Threatened Species - And Make Biodiversity Great Again - The Solutions In The UN IPBES Report

MANY THINGS YOU CAN DO - NATURE USED EARTH SUSTAINABLY FOR MILLIONS OF YEARS - WE CAN TOO

There are many things you can do in your own life to help with climate change and biodiversity loss.

Don’t feel you have to try to do all these things. Any small thing you do helps - it adds up when millions of people are doing the same.

It’s more a case of inspiration, finding new ways forward, to become parrt of that transformative change like teh transition from horse-drawn carriages to cars - this time a transformation just as great, in many parts of your life - but one that leads to a zero carbon generation in the future, and stopping and reversing biodiversity loss and sustainably using our planet.

We can become a sustainable civilization. After all nature managed to use Earth sustainably for millions of years.

• Reduce meat waste - we waste around 20% of meat in our fridges, on our plates etc

• Eat less meat - if you eat meat with every meal, try one day a week without meat. A 20% reduction in meat eating is not only good for health, if we all do this it will be a major reduction in the pressure on tropical rainforests and biodiversity worldwide.

• Eat more meat from chickens, and pigs, or choose meat from animals that are farmed less intensively (the ones raised indoors and fed food that could be eaten by humans have the highest impact on biodiversity loss and global warming)
  
  We use an area equivalent to both Americas to raise animals for meat.
  
  If on average we could reduce meat consumption by a fifth. just 20% (for instance with “Meatless Mondays”) or use lower impact meat, or eliminate that 20% of meat we waste - if we all did it, would save more than the area of the Amazon rainforest which could be used for reafforestation, restoring habitats and however else we please.
  
  This alone would be a huge help when it comes to mitigating global warming and preventing biodiversity loss.

• Reduce food waste of other types. According to IPBES, “Every year, consumers in rich countries waste almost as much food (222 million tonnes) as the entire net food production of sub-Saharan Africa (230 million tonnes).” - that means all of Africa from the Sahara Desert down to South Africa.

• Shop for climate friendly clothing - climate friendly fabrics, such as linen, organic wool and silk. Natural and low impact dyes that use far less water (even white clothes use dyes), clothes that use recycled materials, upcycling, and when you discard your clothes, take them for recycling rather than put them in the trash.
  
  This is taking off in the fashion industry and you can do your part to promote climate friendly clothes - for some examples, see 15 Sustainable Brands That Are Anything But Boring (Elle magazine)

• Switching to peat-free potting compost for your garden or make your own compost.

• Energy saving about the home - LED lighting, choose low energy devices, increase insulation around your hot water cylinder, add draft proofing and loft insulation, install solar panels and if building a new house consider a passive home (lower cost designs now available)

• Reduce car use - walking or cycling to work if feasible, ride sharing, public transport

• If you fly by plane, consider offsetting. As we approach zero emissions, we can reduce aviation emissions to near zero by, for instance, using new forms of aviation fuel, and the rest can be offset by reafforestation and soil improvement. Meanwhile on the personal level the easiest approach is to consider buying carbon offsetting credits.
  
  Also to reduce flying if you can, for instance by using video conferencing or combining flights.
  
  However, although it is likely to be the largest component of your personal emissions if you are a frequent flier - globally, flying accounts for only 2% of total emissions.
  
  Aviation, along with the cement industry, are the two large global industries that have committed to zero emissions by 2050.
  
  So, it is understandable that some activists stop flying as a personal decision, but we do not have to do that globally. We can still fly indeed with increasing numbers of flights. See:
  . Rising to the challenge of zero emissions aviation

• Find space for nature in your garden. One thing you can do in even a fairly small garden is to make a bug hotel for insects

• Give your cat a snazzy colourful collar to stop it killing birds. Bells work too. But make sure the collar unsnaps easily for cat safety.

• Add stickers to your windows to stop birds flying into them and dying

• Share the good science with your friends, and listen to what the scientists themselves say.

• Let others see what you are doing, and talk to them if they ask why - the social network effect

• You do not need to have less children or feel bad about having children - your children are part of this transformative change. Many of our young are in the forefront of this process.
  
  It’s far better to do what you can to help shape their future, so that they can grow up in a world with zero emissions, and pass on the richly biodiverse plaent we all share to the next generation. We do not have a problem of over population, in most of the most developed countries the population is leveling off or decreasing due to prosperity. We need our children!

For more on all this see my:

. Simple lifestyle changes to help reduce global warming and biodiversity loss

EARLIER VERSIONS OF THE MAIN VIDEOS:

Short video

Long video

CONTACT ME VIA ddebunked.org OR EMAIL

You can Direct Message me on Substack - but I check this rarely. Better, email me at support@robertinventor.com

OR contact me at our new forum

https://ddebunked.org

Please do NOT Direct Message me on Facebook any more unless you are already in contact with me.

If already in contact on FB then please understand if I don’t reply for a while except to very urgent messages

Try email or other ways to talk to me until this is sorted out.

For now I only want to talk on FB about how to message me somewhere else.

FOR MORE HELP

To find a debunk see: List of articles in my Debunking Doomsday blog to date See also my Short debunks

Scared and want a story debunked? Post to our forum which is set up for voluntary fact checking.

The forum itself is here:

ddebunked.org

Also do join our forum if you can help with fact checking or to help scared people who are panicking.

Alternatively you can post to our Facebook group. Please look over the group rules before posting or commenting as they help the group to run smoothly

• Facebook group Doomsday Debunked

However I am not able to comment or post there at present because

SEARCH LIST OF DEBUNKS

You can search by title and there’s also an option to search the content of the blog using a google search. Try different terms e.g. Russia, Putin etc as it only searches the title.

CLICK HERE TO SEARCH: List of articles in my Debunking Doomsday blog to date

NEW SHORT DEBUNKS

I do many more fact checks and debunks on our Facebook group than I could ever write up as blog posts. They are shorter and less polished but there is a good chance you may find a short debunk for some recent concern.

I often write them up as “short debunks”

See Latest short debunks for new short debunks

I also tweet the debunks and short debunks to my Blue Sky page here:

. doomsdaydebunks.bsky.social

I do the short debunks more often but they are less polished - they are copies of my longer replies to scared people in the Facebook group.

I go through phases when I do lots of short debunks. Recently, I’ve taken to converting comments in the group into posts in the group that resemble short debunks and most of those haven’t yet been copied over to the wiki.

TIPS FOR DEALING WITH DOOMSDAY FEARS

If suicidal or helping someone suicidal see my:

BLOG: Supporting someone who is suicidal

If you have got scared by any of this, health professionals can help. Many of those affected do get help and find it makes a big difference.

They can’t do fact-checking, don’t expect that of them. But they can do a huge amount to help with the panic, anxiety, maladaptive responses to fear and so on.

Also do remember that therapy is not like physical medicine. The only way a therapist can diagnose or indeed treat you is by talking to you and listening to you. If this dialogue isn’t working for whatever reason, do remember you can always ask to change to another therapist and it doesn’t reflect badly on your current therapist to do this.

Also check out my Seven tips for dealing with doomsday fears based on things that help those scared, including a section about ways that health professionals can help you.

I know that sadly many of the people we help can’t access therapy for one reason or another - usually long waiting lists or the costs.

There is much you can do to help yourself. As well as those seven tips, see my:

BLOG: Doing what matters at times of stress
— new WHO booklet to help with mental health during COVID19 pandemic

BLOG: Breathe in and out slowly and deeply and other ways to calm a panic attack

BLOG: Tips from CBT
— might help some of you to deal with doomsday anxieties

PLEASE DON’T COMMENT HERE WITH POTENTIALLY SCARY QUESTIONS ABOUT OTHER TOPICS - INSTEAD JOIN OUR NEW FORUM ddebunked.org

PLEASE DON’T COMMENT ON THIS POST WITH POTENTIALLY SCARY QUESTIONS ABOUT ANY OTHER TOPIC:

INSTEAD PLEASE COMMENT IN OUR NEW FORUM HERE:

ddebunked.org

It’s just a forum not social media. No age verification. Set up by myself with free open source software Flarum.

For details:

Doomsday Debunked - fact checking stories that scare you

Welcome to our own easy to use new discussion forum at ddebunked.org - voluntary fact checkers helping you if scared of many things - single click to join from Google, X or FB or join via email

We recently opened up our new dedicated Doomsday Debunked discussion forum just for voluntary fact checking for easily scared people. ddebunked.org You are all welcome to join us there…

Read more

a year ago · 3 likes · 1 comment · Robert Walker

If you have any issues joining it do let me know.

Why I ask you to post to our forum with anything scary off topic instead of here

The reason I ask you to post there instead of comment with your concerns about unrelated topics here is that I often can’t respond to comments here for some time. The unanswered comment can scare people who come to this post for help on something else

Also even an answered comment may scare them because they see the comment before my reply.

Do comment here with anything that is on topic for this post, E.g. if you spot any mistakes however small please let me know.

Also, though your first comment should be on topic, it is absolutely fine to digress and go off topic in conversations here in a natural way if that is how the conversation develops.

This is specifically about off topic comments here hat might scare people on a different topic.

PLEASE DON’T TELL A SCARED PERSON THAT THE THING THEY ARE SCARED OF IS TRUE WITHOUT A VERY RELIABLE SOURCE OR IF YOU ARE A VERY RELIABLE SOURCE YOURSELF - AND RESPOND WITH CARE

This is not like a typical post on substack. It is specifically to help people who are very scared with voluntary fact checking. Please no politically motivated exaggerations here. And please be careful, be aware of the context.

We have a rule in the Facebook group and it is the same here.

If you are scared and need help it is absolutely fine to comment about anything to do with the topic of the post that scares you.

But if you are not scared or don’t want help with my voluntary fact checking please don’t comment with any scary material.

If you respond to scared people here please be careful with your sources. Don’t tell them that something they are scared of is true without excellent reliable sources, or if you are a reliable source yourself.

It also matters a lot exactly HOW you respond. E.g. if someone is in an area with a potential for earthquakes there’s a big difference between a reply that talks about the largest earthquake that’s possible there even when based on reliable sources, and says nothing about how to protect themselves and the same reply with a summary and link to measures to take to protect yourself in an earthquake.

The IPCC have very little material on the topic and they neither advocate it nor advocate against it - that’s not how they work. They just present the material on the topic to help us to make decisions ourselves.

IPCC AR6 / WG3 does have some sections on economic degrowth as a way to fight climate change. It has never been done on a country level, and none of the country pledges rely on it, so it is highly theoretical at that level.

However we’ll see that degrowth at a community level is something that is natural, happens already and is widespread especially in indigenous communities.

In this sense then the IPCC and IPBES cover community involvement as a central part of the transformative change that is already happening and needs to be encouraged.

This is an OLD article from Quora - it may need some updates but I’ve copied it over here because I can no longer edit such long articles on Quora.

The main thing that needs updating is that we have continued far more in the direction of the transformative change and sustainable growth than when this was written originally back on June 13th 2022

SOME OF THE MAIN POINTS

• All the IPCC scenarios developed for modelers have GDP growth - projected through to 2100

• Based on these scenarios, IPCC say that climate change action leads to slightly reduced economic growth - but paid for by co-benefits

• Some studies say mitigation of climate change can even increase GDP growth compared to not mitigating it - so long as care is taken that investments in renewables don’t crowd out other investments

• Others say we need negative growth in advanced economies to stay below 2 C - with still positive growth in weaker economies

• The IPCC section is rather short on negative growth and when I look at the degrowth papers they cite - I found puzzling claims - e.g. claim that GDP growth is only possible with energy use growing - when in reality GDP growth in the UK is happening with energy use falling - and claims that we can never rapidly transition to renewables - when many countries already are transitioning rapidly (e.g. Scotland).

• IPCC summary - many countries have decoupled territorial co2 emissions from growth - some but fewer have decoupled consumption-based emissions from GDP

• The idea of a shift of emphasis to economic growth as a tool with good quality of life as the aim - rather than a goal in itself - is central to mainstream thinking on fighting climate change.

Let’s start with an example of what they seem to mean by degrowth.

TIP: the titles of sections in this blog post are like mini-abstracts.

You can get a first impression of this article by reading the headers - then drill down into any section of interest.

EXAMPLE OF CORN HUSKING BEES IN THE USA - COMING TOGETHER AS VOLUNTEERS IN A COMMUNITY TO HUSK CORN IN A FUN OCCASION TO HELP PRODUCERS OF CORN IN THE COMMUNITY

As an example, early pioneers in the USA would do corn husking bees, where communities gather together to have fun as they husk cors during harvest season, competing to see who could hust the most corn.

Not enforced. Nobody policing it. Doing it because they wanted to help their community.

This continues to the present. The early native Indians used far more nutritious corn (maize) than is grown commercially today. The Iroquois white corn project which aims to keep the early nutritious corn going still does corn husking bees - this is an occasion when everyone gathers round to husk the corn and have fun, unpaid voluntary work.

They put out a call for volunteers when they need to husk more corn - which is a long process for the native corn they use. It wouldn’t be economic without voluntary work - but the result is nutritious tasty corn and continuing their native traditions.

The main piece of equipment used in Seneca White Corn processing is a coffee roaster. Unlike yellow corn, Seneca White Corn must be dried, roasted and dehulled (remove the kernels’ indigestible outer covering) before it is used like hominy or be ground into a flour or meal. It’s a labor-intensive processes, much of which is done by hand. The Iroquois White Corn Project processes and sells about 5,000 pounds of corn product a year.

…

They quickly realized they wouldn’t get rich and couldnt’ support a commercial market

“We quickly realized that we weren’t going to get rich doing this,” Jemison says. “In the beginning we were approached by William and Sonoma and other large companies that wanted to carry our products, but we realized that for all of the expense, labor and time of producing that much corn at the price they wanted would hurt what we had set out to do, which is to bring this back to our people.”

It is a labour intensive process that brought together the community and they would share songs and stories as the worked

Preparing the corn for eating is a labor-intensive process, but one that traditionally brought the Seneca community together. Women would sit in a circle as they worked, sharing songs and stories and thereby strengthening and perpetuating Seneca culture.

. The Iroquois White Corn Project | NativeAmerica.travel

Image from

https://bioneers.org/bioneers-iroquois-white-corn-project-brought-native-crops-new-markets-zmaz1801/

This is a photograph of early American settlers doing a similar corn shucking festival.

“The farmers in this section are finishing up wheat sowing, gathering corn and the old time corn shucking is abroad in the land.” This reported in the Mount Airy News in November 1909 which was probably about the time this picture was taken. Farming was labor-intensive work that involved everyone in the family from eldest to youngest. When possible, however, large tasks were done with others in the community and turned into an excuse for socializing, music, and laughter.

. Farming key to community | Mt. Airy News

This is another photograph of a large corn husking celebration

And this is about an entire community of Christian settlers who didn’t use money between themselves for three generations.

See: The Communal, Sometimes Celibate, 19th-Century Ohio Town That Thrived for Three Generations | What It Means to Be American

So on a community level for sure, degrowth works. But nobody has ever scaled this up to a government / state / macroeconomic level.

DEGROWTH ISN’T AUSTERITY - IT’S ABOUT COMMUNITY SPIRIT REPLACING ECONOMIC GROWTH AS A STIMULUS FOR SOLVING OUR PROBLEMS - SOMETHING NEVER ATTEMPTED BEFORE AT COUNTRYWIDE LEVEL - BUT COMMONPLACE AT COMMUNITY LEVEL

By degrowth the IPCC refers to things like coming together as a community to achieve things you can’t do individually, but not as paid work, as something you do with community spirit. This is fun, an enjoyable occasion, completing some project that benefits the entire community but would be calculated as zero GDP, as no money is exchanged.

It’s the idea that human well being can be satisfied without quantifiable economic growth, without money changing hands, when people are motivated to act in other ways.

The idea of countries adopting degrowth strategies is about the idea of scaling up that community spirit to a macroeconomic level to the scale of an entire country. But the writings on this idea are rather academic and they say themselves that it is highly challenging to model as nobody has ever done it before and so they can’t use data from current or previous societies to try to model it.

The idea is that instead of focusing on GDP that we need to focus direction on societal wellbeing. On fulfilling human needs.

This is a quote from a recent academic paper on degrowth cited by the IPCC.

QUOTE Such modelling is highly challenging, partly because a degrowth society would function differently compared to the current society. Thus, model parameters and structures based on past data could no longer be valid. Furthermore, it would need to recognise that GDP is an inadequate indicator for societal wellbeing, at least in affluent countries. Instead, the focus needs to be oriented directly at multidimensional human needs satisfaction9,18,53. This is especially important given that many degrowth proposals include a strengthening of non-monetary work, such as care work and community engagement, as well as decommodification of economic activity towards sharing, gifting and common

. 1.5 °C degrowth scenarios suggest the need for new mitigation pathways - Nature Communications

This is highly theoretical as nobody is proposing this in their pledges and no country has ever experimented with this kind of a major social / political change.

DEGROWTH AT THE MACROECONOMIC LEVEL GETS ONLY A FEW SHORT MENTIONS IN THE IPCC REPORT - AND THEY AREN’T ADVOCATING IT OR ADVOCATING AGAINST IT

The IPCC cover degrowth only briefly, a few sentences in AR6 / WG3. The idea of macroeconomic degrowth is covered in only a few places in a report hundreds of pages long.

Here are the main occurrences - I will just give a brief summary of what each one is about.

As you will see from the examples, the IPCC aren’t advocating degrowth or advocating against it.

This first paragraph presents two views. In the previous page they presented the mainstream view which is that there is a small hit on economic growth from climate mitigation, easily offset by co-benefits.

Then in this page they say that there is a possibility of

• Increased economic activity following mitigation - that it stimulates growth OR

• Large negative economic effects following mitigation.

They then go on to say that some authors say that the transition to a renewable economy will stimulate the economy.

Other authors say that there is no way to stay within 2 C without degrowth, that we need to find a new way to do things not based on economic growth.

Those two views are inconsistent with each other. The IPCC doesn’t attempt to resolve inconsistencies like that, that’s not its role. It just presents the research as it is.

17-59: QUOTE Conversely, several studies find that only a GDP non-growth/degrowth or post-growth approach allow to reach climate stabilization below 2°C (Hardt and O’Neill 2017; D’Alessandro et al. 2020; Hickel and Kallis 2020; Nieto et al. 2020), or to minimize the risks of reliance on high energy-GDP decoupling, large-scale CDR and large-scale renewable energy deployment (Keyßer and Lenzen 2021).

This quote says that reducing waste, reducing emissions, and strong policies of social equity (treating communities and people equally) can lead to a low-carbon transition even without economic growth.

In other words we don’t need economic growth to make the transition in a socially sustainable way - our societies can continue to function as before.

5–32: Such degrowth pathways may be crucial in combining technical 49 feasibility of mitigation with social development goals

This quote says that the degrowth movement has the potential to speed up the transition if large numbers of people take part in it.

17-59: Also, the degrowth movement, with its focus on sustainability over profitability, has the potential to speed up transformations using alternative practices like fostering the exchange of non-monetary goods and services if large numbers of stakeholders want to invest in these areas

3-116 For example, scenarios that include climate change impacts or economic degrowth are not fully represented, as these scenarios, with a few exceptions, were not submitted to the database.

This says that the literature on degrowth argues that prosperity are not immutably linked to economic growth.

The key insight is that we need new holistic ways to measure well-being with at least some measure of decoupling from economic growth. The idea is to choose approaches that maximize synergy between climate mitigation and other sustainable devfelopment goals. Synergy here means they go hand in hand. So the idea is to choose ways to reduce global warming that at the same time make it easier to achieve sustainable development goals such as zero hunger, clean drinking water and sanitation for all, education for all and so on.

Page 1–41: Literature on degrowth, post growth, and post development questions the sustainability and imperative of more growth especially in already industrialised countries and argues that prosperity and the ‘Good Life’ are not immutably tied to economic growth .

The concept of ‘just transition’ also stresses the need to integrate justice concerns so as to not impose hardship on already marginalised populations within and between countries.

The key insight is that pursuing climate goals in the context of sustainable development requires holistic thinking including on how to measure well-being, serious consideration of the notion of ecological limits, at least some level of decoupling and certainly choices and decision-making approaches that exploit and maximise the synergy and minimises the trade-off between climate mitigation and other sustainable development goals. It also requires consideration of equity and justice within and between countries. However, ideas of a synergistic relationship between development and climate mitigation can sometimes offer limited practical guidelines for reconciling the tensions that are often present in actical policy making ).

In all this they just present the research on the topic for others to make use of. Most of the research they summarizes is on the lines that we can solve the problems of climate change using economic growth done carefully with possibly a small hit on economic growth as a result of priorities to stop global warming - though it’s also possible there is no economic hit or even a benefit and that it woudl be paid for by co-benefits if there is any hit. But some think there may be a big hit on economic growth and that we need to shift to new ways of doing things in order to find ways to sustain our societies even without economic growth.

NOT EASY TO WORK OUT WHAT ACADEMICS MEAN BY DEGROWTH

This was a difficult article to write. The problem is that there are people who are very keen on degrowth. But there is very little material in the IPCC report on the topic. I though there was nothing. But they do cover it, so my summary is based on what they say then to look at the papers themselves.

What these academics mean by degrowth is hard to really know as they have no actual examples of a degrowth country to point to. .

Also it seems to be a tiny minority of academics. The IPCC cite many studies and only a handful of their cites are on degrowth.

These academics don’t have any political influence as far as I can tell. There aren’t any examples of even small island states that have tried a degrowth nacroeconomy, whatever that might mean. It remans very abstract at present.

However when it comes to microeconomics then it’s widespread. There are many communities that do things at a community level. Especially indigenous people. At that level there’s much wider support for the idea.

It isn’t about forcing people to volunteer! I don’t think anyone is suggesting that. On a local level, it’s about empowering local communities, not about enforcing anything on them. For indigenous peoples, empowering them to continue managing their forests as they already do

What it means at a country level is harder to know but it’s not communism or enforcing anything on people.

DEGROWTH IS NOT COMMUNISM - BUT SOMETHING COMPLETELY NEW - COMMUNISM IS BASED ON ECONOMIC GROWTH

Degrowth is NOT communism, which relies on putting production in the control of workers but is still about economic growth like all the political and economic systems humanity has tried so far. See my:

. Communism is NOT better for climate change than capitalism and economic growth is NOT bad for the environment or climate change - transformative change is about embracing diversity and finding local solutions

It’s something completely new that nobody has ever tried at the macroeconomic level.

DEGROWTH IN THIS SENSE IS VERY COMMON AT COMMUNITY LEVEL

However it is very common at the community level, particularly in traditional societies, where people come together e.g. to help each other harvest their crops at harvest time - not paying anyone, just doing it to help out a neighbour and later the neighbour helps them when it’s their field’s turn to be harvested.

That is how things used to work in some traditional societies and even today in some indigenous communities they have very little use for moneym=, as a lot of it is people working together out of community spirit, because it’s fun and for mutual help.

NOT SO COMMON TODAY BUT WE STILL HAVE VOLUNTARY WORK WHICH IS SIMILAR

Nowadays many people have that community spirit mainly for their own family and relatives. It used to be a wider network. Back in the nineteenth century say, then many would feel like that about their entire village, say.

We do have it though for some things even today. Mainly as e.g. voluntary projects.

People volunteer to do things to help their community. Example - volunteering for Oxfam or volunteering to help the homeless or conservatrion work. Or soemtimes community work that involves projects such as new renewables organized locally. Or community art events. Or things like a village fete.

This is not meant to present my own views, the examples are to help illustrate ideas. Also it’s not advocacy. The IPCC and IPBES don’t do advocacy, instead they look at things we can do and then look at what the effects would be of doing those things. It’s providing scientific advice for decision makers.

I label the section headers if they are summarizing IPCC research.

IPBES ON INVOLVEMENT OF INDIGENOUS PEOPLES AND LOCAL COMMUNITIES

This isn’t explicitly about degrowth but the way that indigenous peoples and local communities manage their land is often like the Iroquois wheat project, not based on economic growth but on doing things together.

IPBES say that a quarater of the land area of Earth is still managed by indigenous people and this area also includes 35% of the remaining areas with very low human intervention.

IPBES say that global scenarios don’t actually include the ideas of involving indigenous peoples and could benefit from that.

They say that making it possible for them to make decisions about their own lands often enhances their quality of life as well as enhancing nature conservation, restoration and sustainable use of the land.

Indigenous Peoples, Local Communities and Nature

At least a quarter of the global land area is traditionally owned, managed, used or occupied by Indigenous Peoples. These areas include approximately 35% of the area that is formally protected, and approximately 35% of all remaining terrestrial areas with very low human intervention.

Nature managed by Indigenous Peoples and Local Communities is under increasing pressure but is generally declining less rapidly than in other lands – although 72% of local indicators developed and used by Indigenous Peoples and Local Communities show the deterioration of nature that underpins local livelihoods.

The areas of the world projected to experience significant negative effects from global changes in climate, biodiversity, ecosystem functions and nature’s contributions to people are also areas in which large concentrations of Indigenous Peoples and many of the world’s poorest communities reside.

QUOTE Regional and global scenarios currently lack and would benefit from an explicit consideration of the views, perspectives and rights of Indigenous Peoples and Local Communities, their knowledge and understanding of large regions and ecosystems, and their desired future development pathways. Recognition of the knowledge, innovations and practices, institutions and values of Indigenous

Peoples and Local Communities and their inclusion and participation in environmental governance often enhances their quality of life, as well as nature conservation, restoration and sustainable use. Their positive contributions to sustainability can be facilitated through national recognition of land tenure, access and resource rights in accordance with national legislation, the application of free, prior and informed consent, and improved collaboration, fair and equitable sharing of benefits arising from the use, and co-management arrangements with local communities. Media Release: Nature’s Dangerous Decline ‘Unprecedented’; Species Extinction Rates ‘Accelerating’

IPCC SCENARIOS ALL FEATURE ECONOMIC GROWTH - THESE ARE SCENARIOS NOT PREDICTIONS - AND THE SUSTAINABLE DEVELOPMENT PATH SSP1 HAS THE HIGHEST ECONOMIC GROWTH APART FROM SSP5 FOSSIL FUELS OPTIMISM

The 2018 report had an example worst case scenario, where we abandon the Paris agreement before 2020 and don’t try to work together collectively to fight climate change ever, that countries take individual actions starting in the 2030s and we just squeak in at 3 C by 2100 - this wasn’t civilization collapse. This scenario is now impossible.

“Scenario 3 [one possible storyline among worst-case scenarios]:Mitigation: uncoordinated action, major actions late in the 21st century, 3°C of warming in 2100

I go into it here:

I was hoping for an example in AR6 WG2 like the 2018 worst case, but they don’t seem to do that.

They have their “Shared Socioeconomic Pathways” which aren’t meant to be used as predictions of the future but rather as inputs for modelling.

The best of these scenarios in terms of well being in its outcomes is SSP1.

QUOTE SSP1 Sustainability – Taking the Green Road (Low challenges to mitigation and adaptation)

The world shifts gradually, but pervasively, toward a more sustainable path, emphasizing more inclusive development that respects perceived environmental boundaries.

Management of the global commons slowly improves, educational and health investments accelerate the demographic transition, and the emphasis on economic growth shifts toward a broader emphasis on human well-being.

Driven by an increasing commitment to achieving development goals, inequality is reduced both across and within countries. Consumption is oriented toward low material growth and lower resource and energy intensity.

The worst of those is SSP3 regional rivalry similarly to the 2018 worst case example - each can be combined with different mitigation scenarios. Their SSP3-7.0 scenario combines this with all countries abandoning all climate policies through to 2100.

QUOTE A resurgent nationalism, concerns about competitiveness and security, and regional conflicts push countries to increasingly focus on domestic or, at most, regional issues. .... Economic development is slow, consumption is material-intensive, and inequalities persist or worsen over time. Population growth is low in industrialized and high in developing countries. A low international priority for addressing environmental concerns leads to strong environmental degradation in some regions.

It’s not a projection just a scenario.

Historically we have been on SSP2

Middle of the Road (Medium challenges to mitigation and adaptation)

The world follows a path in which social, economic, and technological trends do not shift markedly from historical patterns. Development and income growth proceeds unevenly, with some countries making relatively good progress while others fall short of expectations.

Global and national institutions work toward but make slow progress in achieving sustainable development goals. Environmental systems experience degradation, although there are some improvements and overall the intensity of resource and energy use declines.

Global population growth is moderate and levels off in the second half of the century. Income inequality persists or improves only slowly and challenges to reducing vulnerability to societal and environmental changes remain.

For the other two pathways and more about them, see:

. Explainer: How ‘Shared Socioeconomic Pathways’ explore future climate change - Carbon Brief

We are increasingly moving towards SSP1, in many of our policy decisions, but we aren’t there yet.

ALL THE IPCC SCENARIOS HAVE GDP GROWTH - PROJECTED THROUGH TO 2100

All the IPCC scenarios have GDP growth, projected through to 2100. GDP and population projections from figure 3.4 page 3-19. See image.

All their scenarios see global GDP increase through to 2100. With SSP3 in the AR6 / WG3 the GDP by 2100 is 300 trillion dollars in US 2010 dollars, convert that to 2022 dollars it’s $397.75 trillion - for a population of nearly 13 billion people. GDP and population projections from figure 3.4 page 3-19. See image.

Dollar conversion here: $300 in 2010 → 2022

Compare current global GDP of $87.57 trillion in 2019 (2020 is atypical).

GDP from world bank GDP (current US$)

That’s for a world population of 7.71 billion in 2019 Population

So their GDP per capita in their worst socioeconomic scenario increases from a 2019 figure of $87.57 trillion / $7.71 billion = $11.357 https://www.google.com/searchq=87.57trillion+%2F+7.71+billion

to a 2100 figure of $397.75 trillion / 13 billion = $30,596 in 2022 USD

So global GDP per capita nearly triples by 2100 in their worst case of regional rivalry. 397.75trillion / 13 billion

That’s a global average GDP per capita similar to Puerto Rico. Higher than Malta, Cyprus, or Spain. The current average is between the GDP of China and of Costa Rica.

. GDP per capita (current US$)

In their sustainable development path SSP1 it increases to just under $600 trillion, call it $590 trillion, so twice as much by 2100 and with a lower population by 2100 of less than 7 billion, looks like about 6.9 billion (in the sustainable development path low income countries like Africa are much more prosperous by 2100 and so has a much lower population as its birth rate falls fast).

Convert to 2022 dollars and that’s 782.25 trillion $590 in 2010 → 2022

For a world population of 6.9 billion so that works out at $113,370 GDP per capita 782.25 trillion / 6.9 billion

That’s a global average GDP per capita by 2100 between present day Bermuda and Luxembourg.

IPCC DISCUSSION OF OTHER POSSIBILITIES OUTSIDE OF THE RANGE OF THEIR FIVE SCENARIOS SSP1 TO SSP5 - THERE MAY BE A POSSIBILITY OF MUCH FASTER ECONOMIC GROWTH ON THE SUSTAINABLE DEVELOPMENT PATHS THAN USUALLY ASSUMED, WHILE OTHERS THINK THAT ZERO OR EVEN NEGATIVE GROWTH COULD HAPPEN

They say that most economic projections assume growth rates return to normal after the COVID pandemic.

QUOTE Economic growth is even more uncertain than the population projections (Figure 3.9). The average growth rate of GDP was about 2.8% per year (constant USD) in the 1990-2019 period (The World Bank 2021). In 2020, the Covid-19 crisis resulted in a considerable drop in GDP (estimated around 4- 5%). After a recovery period, most economic projections assume growth rates to converge back to previous projections, although at a lower level

In the long term what happens about economic growth depends on political stability, and technoological progress, including the extent to which countries can catch up.

Scenarios outside the range they look at are possible including economic decline, or much faster economic growth than they considered.

QUOTE In the long-term, assumptions on future growth relate to political stability, the role of the progress of the technology frontier and the degree to which countries can catch up . The SSP scenarios cover an extensive range, with low per capita growth in SSP3 and SSP4 (mostly in developing countries) and rapid growth in SSP1 and SSP5. At the same, however, also scenarios outside the range have some plausibility – including the option of economic decline or much faster economic development

[More on this in the following sections]

The current long term predictions are reasonably consistent with SSP2.

QUOTE The OECD long-term projection is at the global level reasonably consistent with SSP2. Equally important economic parameters include income distribution (inequity) and the type of growth (structural change, i.e., services vs manufacturing industries). Some projections (like SSP1) show a considerable convergence of income levels within and across countries, while in other projections, this does not occur (e.g., SSP3).

QUOTE Most scenarios reflect the suggested inverse relationship between the assumed growth rate for income and population growth (Figure 3.9e). SSP1 and SSP5 represent examples of scenarios with relatively low population increase and relatively high-income increase over the century. SSP3 represents an example of the opposite – while SSP2 and SSP4 are placed more in the middle.

The scenarios usually don’t take account of the impact of climate on growth. Some literature emphasizes the possibility of incomes stabilizing or even declining in developed countries and argue that this might be better or even needed than for the income to keep growing indefinitely.

The Integrated Assessment Models however normally assume that decarbonization can be combined with economic growth with changes in technology, lifestyle and how the economy works (that would include things like the shifts of harmful subsidies to subsidize activities that benefit nature services).

QUOTE Nearly all scenarios assessed here do not account for climate impacts on growth (mostly for methodological reasons). As discussed in Section 3.5 these impacts can be considerable. An emerging area of literature emphasises the possibility of stabilisation (or even decline) of income levels in developed countries, arguing that such a trend would be preferred or even needed for environmental reasons (see also Chapter 5). Such scenarios are not common among IAM outcomes, that are more commonly based on the idea that decarbonisation can be combined with economic growth by a combination of technology, lifestyle and structural economic changes. Still, such scenarios could resuln a dramatic reduction of energy and resource consumption (see further)

SSP5 ACHIEVES HIGHEST ECONOMIC GROWTH OF ALL - OPTIMISTIC ABOUT TECHNOLOGY TO SOLVE PROBLEMS INCLUDING GEOENGINEERING - BUT THIS ONE IS RISKY AS IT IS HARD TO STAY WITHIN 1.5 OR 2 C AND IT LEADS TO MUCH MORE HARM ON BIODIVERSITY AND NATURE SERVICES

You might wonder about SSP5 which is way to the right with the highest economic growth of all. This is an optimistic fossil fuel powered scenario - but one where it is very hard to stay within 1.5 C or 2 C, one that relies a lot on technological fixes such as geoengineering - and high risk of destruction of Nature services and shifts to new ecosystems. To put in context at above 3 C coral reefs are no longer possible and transition to sponges or seaweeds or a mix of both. The result is still a very habitable world, it’s not talking about a world without nature - but one where the ecossytems we have transition to new ones and species find it hard to adapt and many would go extinct.

See my Coral reefs effects MID EDIT - SECOND HALF

Fossil-fueled Development – Taking the Highway (High challenges to mitigation, low challenges to adaptation)
This world places increasing faith in competitive markets, innovation and participatory societies to produce rapid technological progress and development of human capital as the path to sustainable development. Global markets are increasingly integrated. There are also strong investments in health, education, and institutions to enhance human and social capital. At the same time, the push for economic and social development is coupled with the exploitation of abundant fossil fuel resources and the adoption of resource and energy intensive lifestyles around the world. All these factors lead to rapid growth of the global economy, while global population peaks and declines in the 21st century. Local environmental problems like air pollution are successfully managed. There is faith in the ability to effectively manage social and ecological systems, including by geo-engineering if necessary.

SSP5 has the highest growth but risks a temperature increase of up to 5 C with all the problems that would entail.

In SSP5, emissions rise too far and fall too slowly to meet the Paris targets without large amounts of negative emissions in the latter part of the century, to offset slower near-term emissions reductions. Three out of four SSP5 models were able to develop possible scenarios to achieve RCP2.6 targets of limiting warming below 2C, while only two out of four were able to find a way to reach RCP1.9 and limit warming to below 1.5C.

. Explainer: How ‘Shared Socioeconomic Pathways’ explore future climate change - Carbon Brief

Bear in mind all of these are scenarios, not predictions.

CLIMATE SCENARIO CATEGORIES C1 = 1.5 C WITH NO OVERSHOOT, C2 = 1.5 C WITH OVERSHOOT, C3 = 2 C WITH 67% PROBABILITY, C4 = 2 C WITH 50% PROBABILITY, C5 = 2.5 C, C6 = 3 C, C7 = 4 C AND C8 = > 4 C

This is about the various warming scenarios used in the IPCC report and is for a reference for the next section.

• Category C1 comprises modelled scenarios that limit warming to 1.5°C in 2100 with a likelihood of greater than 50%, and reach or exceed warming of 1.5°C during the 21st century with a likelihood of 67% or less.
  
  In this report, these scenarios are referred to as scenarios that limit warming to 1.5°C (>50%) with no or limited overshoot. Limited overshoot refers to exceeding 1.5°C global warming by up to about 0.1°C and for up to several decades. [FOOTNOTE 49]

• Category C2 comprises modelled scenarios that limit warming to 1.5°C in 2100 with a likelihood of greater than 50%, and exceed warming of 1.5°C during the 21st century with a likelihood of greater than 67%. In this report, these scenarios are also referred to as scenarios APPROVED Summary for Policymakers IPCC AR6 WG III Subject to copyedit SPM-28 Total pages: 63 that return warming to 1.5°C (>50%) after a high overshoot. High overshoot refers to temporarily exceeding 1.5°C global warming by 0.1-0.3°C for up to several decades.

• Category C3 comprises modelled scenarios that limit peak warming to 2°C throughout the 21st century with a likelihood of greater than 67%. In this report, these scenarios are also referred to as scenarios that limit warming to 2°C (>67%).

• Categories C4-C7 comprise modelled scenarios that limit warming to 2°C, 2.5°C, 3°C, 4°C, respectively, throughout the 21st century with a likelihood of greater than 50%. In some scenarios in C4 and many scenarios in C5-C7, warming continues beyond the 21st century.

• Category C8 comprises modelled scenarios that exceed warming of 4°C during the 21st century with a likelihood of 50% or greater. In these scenarios warming continues to rise beyond the 21st century.

IPCC SAY THAT CLIMATE CHANGE ACTION LEADS TO SLIGHTLY REDUCED ECONOMIC GROWTH - BUT PAID FOR BY CO-BENEFITS

Economically aiming for 1.5 C is better than 2 C. The extra ecomomic costs are more than paid for by the benefits.

A policy of 1.5 C without overshoot reduces global GDP annual growth by 0.09% to 0.14% and more than pays for itself with cobenefits

Chart of world GDP growth from the world bank GDP growth (annual %)

For 1.5 C with overshoot (Scnario C1) it’s between 0.05% and 0.09% a year and for a 50% chance of exceeding 2 C it’s between 0.03 and 0.07% per year.

QUOTE C.12.2 The aggregate effects of climate change mitigation on global GDP are small compared to global projected GDP growth in assessed modelled global scenarios that quantify the macroeconomic implications of climate change mitigation, but that do not account for damages from climate change nor adaptation costs (high confidence). For example, compared to pathways that assume the continuation of policies implemented by the end of 2020, assessed global GDP reached in 2050 is reduced by 1.3–2.7% in modelled pathways assuming coordinated global action starting between now and 2025 at the latest to limit warming to 2°C (>67%). The corresponding average reduction in annual global GDP growth over 2020-2050 is 0.04–0.09 percentage points. In assessed modelled pathways, regardless of the level of mitigation action, global GDP is projected to at least double (increase by at least 100%) over 2020-2050.

For modelled global pathways in other temperature categories, the reductions in global GDP in 2050 compared to pathways that assume the continuation of policies implemented by the end of 2020 are as follows: 2.6 - 4.2% (C1), 1.6 - 2.8% (C2), 0.8 - 2.1% (C4), 0.5 - 1.2% (C5). The corresponding reductions in average annual global GDP growth over 2020-2050, in percentage points, are as follows: 0.09 - 0.14 (C1), 0.05 - 0.09 (C2), 0.03 - 0.07 (C4), 0.02 - 0.04 (C5) [FOOTNOTE 67]. There are large variations in the modelled effects of mitigation on GDP across regions, depending notably on economic structure, regional emissions reductions, policy design and level of international cooperation [FOOTNOTE 68] (high confidence). Country level studies also show large variations in the effect of mitigation on GDP depending notably on the level of mitigation and on the way it is achieved (high confidence). Macroeconomic implications of mitigation co-benefits and trade-offs are not quantified comprehensively across the above scenarios and depend strongly on mitigation strategies (high confidence). {3.6, 4.2, Box TS.7, Annex III I.2, I.9, I.10 and II.3}

The costs on GDP growth don’t necessarily mean actual reduced GDP because these calculations don’t take account of the effect of damage on economic growth from climate change if it is not mitigated.

In the majority of studies that focus on the economic effects of mitigation without accounting for climate damages, global economic growth and consumption growth is reduced compared to baseline scenarios (that omit damages from climate change), but mitigation pathways do not represent an absolute decrease of economic activity el (Figure 3.34, panels b and c).

ECONOMIC BENEFITS OUTWEIGH IMMEDIATE COSTS AT MEDIUM TO HIGH RANGE OF ASSESSED IMPACTS FOR 2 C AND MITIGATION IS A WELFARE ENHANCING STRATEGY (HIGH CONFIDENCE)

TS-48

QUOTE The global benefits of pathways likely to limit warming to 2°C outweigh global mitigation costs over the 21st 5 century, if aggregated economic impacts of climate change are at the moderate to high end of the assessed range, and a weight consistent with economic theory is given to economic impacts over the long-term. This holds true even without accounting for benefits in other sustainable development dimensions or non-market damages from climate change (medium confidence).

The aggregate global economic repercussions of mitigation pathways include: the macroeconomic impacts of investments in low-carbon solutions and structural changes away from emitting activities; co-benefits and adverse side effects of mitigation; avoided climate change impacts; and, reduced adaptation costs. Existing quantifications of the global aggregate economic impacts show a strong dependence on socioeconomic development conditions, as these shape exposure and vulnerability and adaptation opportunities and responses. Avoided impacts for poorer households and poorer countries represent a smaller share in aggregate economic quantifications expressed in GDP or monetary terms, whereas their well-being and welfare effects are comparatively larger.

When aggregate economic benefits from avoided climate change impacts are accounted for, mitigation is a welfare enhancing strategy (high confidence). {3.6.2} The economic benefits on human health from air quality improvement arising from mitigation action can be of the same order of magnitude as mitigation costs, and potentially even larger (medium confidence). {3.6.3}

HIGH CONFIDENCE THAT SOME SECTORS AND REGIONS WILL SEE SUPERIOR ECONOMIC BENEFITS OF ENERGY TRANSITION TO RENEWABLES COMPARED TO FOSSIL FUELS S - AND THIS IMPROVES AS TECHNOLOGY IMPROVES

TS-54

QUOTE A low-carbon energy transition will shift investment patterns and create new economic opportunities (high confidence). Total energy investment needs will rise, relative to today, over the next decades, if it is to be likely that warming is limited to 2°C, or if warming is limited to 1.5°C with no or limited overshoot. These increases will be far less pronounced, however, than the reallocations of investment flows that are anticipated across sub-sectors, namely from fossil fuels (extraction, conversion, and electricity generation) without CCS and toward renewables, nuclear power, CCS, electricity networks and storage, and end-use energy efficiency. A significant and growing share of investments between now and 2050 will be made in emerging economies, particularly in Asia. {6.7}

TS-57

QUOTE The economic outcomes of low-carbon transitions in some sectors and regions may be on par with, or superior to those of an emissions-intensive future (high confidence). Cost reductions in key technologies, particularly in electricity and light-duty transport, have increased the economic attractiveness of near-term low-carbon transitions. Long-term mitigation costs are not well understood and depend on policy design and implementation, and the future costs and availability of technologies. Advances in low-carbon energy resources and carriers such as next-generation biofuels, hydrogen produced from electrolysis, synthetic fuels, and carbon-neutral ammonia would substantially improve the economics of net zero energy systems (medium confidence). {6.4, 6.7}